Tariffs: persist tariff values in 15min intervals

[naltatis]

builts upon #27780

Sketch for the dedicated tariff log schema discussed in the PR review. Instead of logging outside temperature like a regular meter, tariff values get their own lean table.

• new tariffs table: one row per quarter hour with ts, grid, feedin, co2 and temperature columns, unconfigured tariffs stay NULL
• values are read at the exact 15min boundary via tariff.At, so the stored timestamp is always x:00/x:15/x:30/x:45
• persistence is driven by the existing site update loop like the meter collectors, the partial boot slot is skipped
• reverts the temperature-as-meter logging (soc_temp collector path), UI publishing is unchanged

[sourcery-ai]

Hey - I've found 1 issue, and left some high level feedback:

• PersistTariffs assumes the caller always passes a 15-minute boundary; consider validating the timestamp alignment inside the function to guard against accidental misuse.
• PersistTariffs uses multiple *float64 parameters in a fixed order; consider introducing a struct or named type to make call sites less error-prone and more self-documenting.
• The OnConflict DoNothing behavior means updated tariff values for an existing slot are silently ignored; if later corrections are expected, consider using an upsert or explicit update instead.

Prompt for AI Agents

Please address the comments from this code review:

## Overall Comments
- PersistTariffs assumes the caller always passes a 15-minute boundary; consider validating the timestamp alignment inside the function to guard against accidental misuse.
- PersistTariffs uses multiple *float64 parameters in a fixed order; consider introducing a struct or named type to make call sites less error-prone and more self-documenting.
- The OnConflict DoNothing behavior means updated tariff values for an existing slot are silently ignored; if later corrections are expected, consider using an upsert or explicit update instead.

## Individual Comments

### Comment 1
<location path="core/metrics/tariffs_test.go" line_range="11" />
<code_context>
+	"github.com/stretchr/testify/require"
+)
+
+func TestPersistTariffs(t *testing.T) {
+	require.NoError(t, db.NewInstance("sqlite", ":memory:"))
+	require.NoError(t, db.Instance.AutoMigrate(new(tariffValue)))
</code_context>
<issue_to_address>
**suggestion (testing):** Extend `TestPersistTariffs` to cover additional non-nil combinations and temperature/feedin fields

To strengthen coverage, add subtests (e.g. via `t.Run`) that exercise cases where:
- `feedin` is non-nil and persisted correctly.
- `temperature` is non-nil and persisted correctly.
- Mixed combinations (e.g. `grid=nil` with non-nil `co2` and `temperature`) confirm each column is mapped and stored independently.
This ensures all `tariffs` columns are covered by tests and field-to-column mapping is fully validated.

Suggested implementation:

```golang
func TestPersistTariffs(t *testing.T) {
	require.NoError(t, db.NewInstance("sqlite", ":memory:"))
	require.NoError(t, db.Instance.AutoMigrate(new(tariffValue)))

	t.Run("omit nil values", func(t *testing.T) {
		// ensure a clean table for this case
		require.NoError(t, db.Instance.Exec("DELETE FROM tariff_values").Error)

		slot := time.Date(2026, 4, 15, 16, 15, 0, 0, time.UTC)
		grid, co2 := 0.3, 250.0

		// nil values omitted
		require.NoError(t, PersistTariffs(slot, &grid, nil, &co2, nil))

		var res tariffValue
		require.NoError(t, db.Instance.First(&res).Error)
		require.Equal(t, slot.Unix(), res.Timestamp)

		// verify non-nil fields are persisted and mapped correctly
		require.NotNil(t, res.Grid)
		require.Equal(t, grid, *res.Grid)

		require.NotNil(t, res.Co2)
		require.Equal(t, co2, *res.Co2)

		// verify nil fields are not persisted
		require.Nil(t, res.FeedIn)
		require.Nil(t, res.Temperature)
	})

	t.Run("persist feedin", func(t *testing.T) {
		require.NoError(t, db.Instance.Exec("DELETE FROM tariff_values").Error)

		slot := time.Date(2026, 4, 15, 17, 0, 0, 0, time.UTC)
		grid, feedin := 0.25, 0.18

		require.NoError(t, PersistTariffs(slot, &grid, &feedin, nil, nil))

		var res tariffValue
		require.NoError(t, db.Instance.First(&res).Error)
		require.Equal(t, slot.Unix(), res.Timestamp)

		require.NotNil(t, res.Grid)
		require.Equal(t, grid, *res.Grid)

		require.NotNil(t, res.FeedIn)
		require.Equal(t, feedin, *res.FeedIn)

		require.Nil(t, res.Co2)
		require.Nil(t, res.Temperature)
	})

	t.Run("persist temperature", func(t *testing.T) {
		require.NoError(t, db.Instance.Exec("DELETE FROM tariff_values").Error)

		slot := time.Date(2026, 4, 15, 17, 30, 0, 0, time.UTC)
		co2, temperature := 260.0, 22.5

		require.NoError(t, PersistTariffs(slot, nil, nil, &co2, &temperature))

		var res tariffValue
		require.NoError(t, db.Instance.First(&res).Error)
		require.Equal(t, slot.Unix(), res.Timestamp)

		require.Nil(t, res.Grid)
		require.Nil(t, res.FeedIn)

		require.NotNil(t, res.Co2)
		require.Equal(t, co2, *res.Co2)

		require.NotNil(t, res.Temperature)
		require.Equal(t, temperature, *res.Temperature)
	})

	t.Run("mixed combinations", func(t *testing.T) {
		require.NoError(t, db.Instance.Exec("DELETE FROM tariff_values").Error)

		slot := time.Date(2026, 4, 15, 18, 0, 0, 0, time.UTC)
		co2, temperature, feedin := 275.0, 19.3, 0.21

		// grid=nil; co2 + temperature + feedin non-nil
		require.NoError(t, PersistTariffs(slot, nil, &feedin, &co2, &temperature))

		var res tariffValue
		require.NoError(t, db.Instance.First(&res).Error)
		require.Equal(t, slot.Unix(), res.Timestamp)

		// each column mapped and stored independently
		require.Nil(t, res.Grid)

		require.NotNil(t, res.FeedIn)
		require.Equal(t, feedin, *res.FeedIn)

		require.NotNil(t, res.Co2)
		require.Equal(t, co2, *res.Co2)

		require.NotNil(t, res.Temperature)
		require.Equal(t, temperature, *res.Temperature)
	})

```

These edits assume the following:
- `tariffValue` has pointer fields `Grid`, `FeedIn`, `Co2`, and `Temperature` of type `*float64`.
- The underlying table name is `tariff_values` (used in the `DELETE` statements).

If the struct uses different field names or types (e.g. non-pointer floats) or a different table name, you should:
1. Adjust the assertions to match the actual field names and types (e.g. drop `NotNil`/`Nil` checks for non-pointer fields and use direct equality).
2. Update the `DELETE FROM tariff_values` SQL to use the correct table name for `tariffValue`.
</issue_to_address>

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}

[sourcery-ai]

Hey - I've reviewed your changes and they look great!

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}