fix: MiniMax max_tokens forwarding + test_learn_episodic_mirror hygiene

[diazMelgarejo]

Two independent, unrelated fixes found while working on #64 in this same repo

1. harness/llm.py — MiniMax OpenAI wire drops max_tokens

call_model()'s MiniMax path forwards max_tokens correctly on the Anthropic wire but silently drops it entirely on the OpenAI-compatible wire — chat.completions.create() passes model/temperature/messages and nothing for output length. Every MiniMax call through that wire has been unbounded regardless of what the caller asked for.

MiniMax's OpenAI-compatible Chat Completions API deprecated bare max_tokens in favor of max_completion_tokens; this adds it. Extended test_call_model_openai_wire_global with an assertion on the new kwarg — verified it fails with KeyError against the pre-fix code and passes with the fix.

2. tools/test_learn_episodic_mirror.py — module-stub leak + missing encoding

_load_learn() stubs sys.modules["text"]/["cluster"] to isolate learn.py from its two sibling modules for the test, but never removed the stubs afterward — they stayed process-wide for the rest of the test run. This isn't theoretical: ran it directly and confirmed both names remain in sys.modules after one call, meaning any later test or import in the same process would silently get the throwaway lambda stand-ins instead of the real modules. Now saved and restored (or removed, if there was nothing there before) in a finally block around exec_module().

Also added encoding="utf-8" to the one read_text() call in this file using the platform default, matching the convention this file already follows everywhere else (_episodic(), learn.py itself).

Deliberately not touched: test_append_mirror_fails_open_on_write_error. _append_episodic_mirror()'s fail-open behavior on a write error is this file's own explicit, documented design choice — its docstring says so directly, matching _lesson_already_appended's read-only fail-open posture elsewhere in the same file. That's a design decision, not a bug, and this PR leaves it exactly as-is.

Tests

pytest tests/test_llm_provider.py — 8 passed. python3 -m unittest test_learn_episodic_mirror (from .agent/tools/) — 3 passed. Both fixes verified in both directions (fails pre-fix, passes post-fix) before opening this PR.

Scope

2 commits, one per fix, each independently revertable. No schema changes, no new dependencies.

Note

Fix max_tokens forwarding for MiniMax and fix module stub leaks in episodic mirror tests

• Adds max_completion_tokens to the OpenAI-compatible wire call in _call_minimax, so the caller's max_tokens value is now respected by the MiniMax provider.
• Fixes stub leakage in _load_learn: text and cluster modules injected into sys.modules are now restored (or removed) in a finally block after exec_module completes.
• Adds a test assertion in test_call_model_openai_wire_global verifying max_completion_tokens=4096 is passed through.

^{Macroscope summarized 20441c2.}

[diazMelgarejo]

Related: #64 (the loop-event-export allowlist fix this PR's fixes were found alongside).