Capture starting Plating for enemies that apply it in a helper

[ptrlrd]

Bug report (via Knowledge Demon): Lagavulin Matriarch's page shows no indication of its starting Plating.

Root cause

The monster parser scans AfterAddedToRoom() for PowerCmd.Apply<...> calls to build innate_powers. Frog Knight applies its Plating inline there, so it was captured. Lagavulin Matriarch delegates to a private helper — AfterAddedToRoom() calls await Sleep(), and the Plating is applied inside Sleep() — so the scan, which only looked at the literal AfterAddedToRoom body, missed it. (So the report's "Frog Knight also has no indication" was already fixed; the real gap was Lagavulin.)

The frontend already renders innate_powers — the field was just null.

Fix

The parser now follows local same-class method calls from AfterAddedToRoom (await Foo(...) where Foo is a method defined in this class, not base.X / Cmd.X) into their bodies before scanning, so powers applied one hop away are caught. Bounded with a visited set.

Validation

I ran the old and new parser against the same decompiled source and diffed innate_powers across every monster:

• Exactly one monster changes: LAGAVULIN_MATRIARCH → Plating 12 and Asleep 3 (both genuinely applied at combat start in Sleep(); both are real catalog powers that render).
• Zero regressions — every other monster (including the 45 that already had innate_powers) is byte-identical.

Data regenerated for all 14 languages on both the stable and beta channels (the value is language-independent — power_id/amount only). One parser file + 28 monsters.json.