[WIP] Add high-ROI coverage targets including tests and CI visibility

[Copilot]

Thanks for the feedback on #1347. I've created this new PR, which merges into #1347, to address your comment. I will work on the changes and keep this PR's description up to date as I make progress.

Original PR: #1347
Triggering comment (#1347 (comment)):

@copilot I found the core issue: this job is running a very broad test matrix, but the dominant failures are from missing optional dependencies (pyarrow) and likely test/env mismatch rather than one localized regression.

Most actionable fix (likely unblocks dozens of failures)

Your logs show repeated hard failures like:

• Py3plexFormatError: pyarrow is required for Apache Arrow format...
• Py3plexIOError: PyArrow is required for Parquet support...

These should not fail in a default test environment unless:

1. CI is expected to include Arrow/Parquet extras but doesn’t install them, or
2. tests are not guarded with importorskip("pyarrow").

✅ Solution

1) Update workflow install step to include test extras + arrow

In .github/workflows/tests.yml (job that runs pytest), install:

- name: Install dependencies
  run: |
    python -m pip install --upgrade pip
    pip install -e ".[test,arrow]"

If your extras differ, equivalent explicit fallback:

pip install -e . pytest pyarrow

2) Guard Arrow/Parquet tests if optional dependency is truly optional

At top of Arrow/Parquet test modules (e.g. tests/test_roundtrip_invariants.py sections for Arrow/Parquet):

import pytest

pyarrow = pytest.importorskip("pyarrow", reason="pyarrow optional dependency not installed")

This ensures default CI won’t fail if pyarrow is intentionally not installed.

---

Secondary failures (after env fix)

You still have many genuine code/test contract mismatches. Prioritize these clusters:

1. DSL Query API regressions

   • AttributeError: 'Query' object has no attribute 'summary'
   • 'list' object has no attribute '_to_ast'
   • 'dict' object has no attribute 'count' / 'to_pandas'

   These indicate return-type/API changes not reflected in tests (or accidental breaking change in builder internals).
   Fix direction: restore compatibility methods (summary, AST conversions), or update tests + adapters consistently.

2. Semiring validation error semantics changed

   • tests expect commutativity / distributivity, got identity_plus / associativity.

   Fix direction: either restore prior validation ordering/messages or relax tests to assert error code/category instead of exact property order.

3. UQ reducer keying changed

   • tests expect reducer names (Reducer1, TestReducer), results keyed by 'MockClass'.

   Fix direction: key outputs by reducer instance name/label (or update tests to new canonical key, consistently across provenance + outputs).

4. Temporal view null handling

   • AttributeError: 'NoneType' object has no attribute 'edges' in empty network test.

   Fix direction: add null-safe path for empty base graph in temporal view constructor/accessors.

---

Minimal PR patch recommendation

1. First PR (stabilize CI):
   • add pyarrow to CI install OR skip optional tests when missing.
2. Second PR (API compatibility pass):
   • fix Query .summary(), AST layer list handling, consistent result wrapper (count, to_pandas).
3. Third PR (behavioral semantics):
   • semiring/UQ expectation alignment.

This sequence should convert the current “130 failed / 20 errors” into a much smaller, tractable set quickly.