REG-054 — FIXED: aegis/core/mmr.py's module-level comment corrected to
match the actual default set by a prior session's REG P1-2 work.

=== The stale claim ===

Before (aegis/core/mmr.py, lines ~70-71):
    # ``CryptographicAuditLedger`` now exposes the choice through
    # ``mmr_hash_scheme``, still defaulting to v1. Two prerequisites blocked that
    # and both are met:

This was accurate when written (both prerequisites for switching the
default were met, but the switch had not yet happened). A later session's
work (task history: "P1-2: default new chains to MMR v2, adopt the WAL's
scheme for existing ones") changed the actual default, but this comment
was never updated to reflect it — leaving stale in-code documentation
directly contradicting the current behavior.

=== Confirmed the actual current default (source, not memory) ===

$ sed -n '410,421p' aegis/config.py
    mmr_hash_scheme: Literal["auto", "v1-asciihex", "v2-binary-domain-separated"] = Field(
        default="auto",
        description=(
            "MMR hash construction for the audit ledger. 'auto' (the default) starts a NEW "
            "chain on 'v2-binary-domain-separated', which applies RFC 6962 domain tags and so "
            "prevents the leaf/interior-node type confusion v1 admits, and reopens an EXISTING "
            "chain under whichever scheme its WAL recorded. ..."
        ),
    )

Confirmed: `auto` (the config default) starts NEW chains on v2, not v1.
`aegis/core/crypto_audit.py:792-794` implements this:
`self.mmr_hash_scheme = (mmr_hash_scheme if self._scheme_pinned else
MMR_NEW_CHAIN_DEFAULT_SCHEME)`.

=== Fix ===

Rewrote the comment block (aegis/core/mmr.py, module-level, the "Hash
schemes" section) to state the actual current behavior: `auto` starts a
new chain on v2 and reopens an existing chain under whichever scheme its
WAL recorded, so a v1 chain in the field stays v1 (no in-place upgrade)
and only a never-opened chain starts on v2. Also removed "and remains the
default" from the preceding sentence about v1 being retained, since v1
remains the scheme for *existing* chains but is no longer *the* default
for new ones — that distinction is exactly what was previously blurred.

Doc-only change to a code comment; no behavior touched.

=== Tests ===

$ .venv/bin/python -m pytest tests/test_mmr_v2_migration.py tests/ -k "mmr" -q
378 passed, 6631 deselected in 6.03s

No test asserts on the comment text itself (confirmed no grep hits for the
old wording anywhere in docs/ either — REG-054's own registry row was the
only citation of the stale line).
