fix(mass_transfer): deprecate silently-ignored m_slope arg in ntu_og

[defnalk]

Fixes #5

Summary

• `mass_transfer.ntu_og` accepted an `m_slope` parameter that the docstring claimed enabled an equilibrium-corrected calculation, but the function body never referenced it. Every call returned the dilute-limit `ln(Y_bottom / Y_top)` regardless of `m_slope`. Callers using the supposed rigorous mode were silently getting the optimistic dilute approximation on real plant data.
• This PR is the semver-clean v0.1.x fix: non-zero `m_slope` now emits a `DeprecationWarning` pointing at ntu_og silently ignores m_slope parameter #5 and stating that the parameter will be removed in v0.2.0. The `m_slope=0.0` default path is unchanged and warning-free.
• A proper Colburn-equation entry point taking the absorption factor explicitly is planned for v0.2.0 (tracked in ntu_og silently ignores m_slope parameter #5).

Why a deprecation, not an immediate fix?

The current signature has no way to compute Colburn's equation — it would also need the L/G molar flows. Implementing the rigorous form requires a breaking signature change, which belongs in v0.2.0 under semver. Until then, the honest move is to surface the bug loudly without breaking existing scripts that pass `m_slope=0`.

Before / after

```python

BEFORE — silent dilute-limit answer regardless of m_slope

ntu_og(0.14, 0.02, m_slope=0.85)
1.9676... # identical to ntu_og(0.14, 0.02), no warning
```

```python

AFTER — same value, but the user is told

ntu_og(0.14, 0.02, m_slope=0.85)
DeprecationWarning: ntu_og(m_slope=...) has never been implemented and is silently
ignored — the function always returns the dilute-limit ln(Y_bottom / Y_top). The
parameter will be removed in meapy v0.2.0; until then, calling with a non-zero
m_slope produces the same result as calling with m_slope=0. See
#5 for the planned Colburn-based replacement API.
1.9676...
```

The default `m_slope=0` call path is unchanged:
```python

ntu_og(0.14, 0.02) # no warning, same value as before
1.9676...
```

Testing notes

Two new tests pin the behaviour in `tests/unit/test_mass_transfer.py::TestNTUOG`:

• `test_nonzero_m_slope_warns_and_returns_dilute_limit` — asserts the warning fires and the returned value is bit-identical to the `m_slope=0` case.
• `test_zero_m_slope_does_not_warn` — runs both `ntu_og(0.14, 0.02)` and `ntu_og(0.14, 0.02, m_slope=0.0)` under `warnings.simplefilter("error")` to guarantee the default path stays clean.

Local run:
```
$ pytest tests/unit/test_mass_transfer.py -q
40 passed in 0.13s
```

Full unit suite (124 + 2 new + the four other commits already on main):
```
$ pytest tests/ -q
130 passed
```

Out of scope (follow-ups)

• The Colburn replacement API itself — will land in v0.2.0 per ntu_og silently ignores m_slope parameter #5.
• CHANGELOG entry — included in a separate `docs(changelog)` commit on main since several other audit commits also need entries.

🤖 Generated with Claude Code