fix(rawir): ACF rows leave under field names (0.6.1)

[ABB65]

ACF returns unformatted repeater/group/flexible rows keyed by sub-field key (field_5f3…); REST gives names. Source::acf_value now renames row keys to sub-field names recursively (repeater, group, flexible, nested). A field_* key no sub-field owns stays as is and is reported as acf-key-unmapped. Tests: repeater, group+nested, flexible, unknown key; integration suite 388 checks pass. Version and changelog 0.6.1.

[ABB65]

ONAY on ea3b6a8 (t2).

Checked against the code, not the description:

• Recursion: acf_value renames row keys to sub-field names and recurses through each sub-field's own definition, so repeater → group → repeater and flexible rows are covered. Tests cover repeater (keyed + already-named rows), group with a nested repeater, and flexible. The flexible acf_fc_layout marker is not a field_* key and is kept uncounted.
• Unresolved keys: a field_* key no sub-field owns keeps its cell and its key (no invented name) and adds acf-key-unmapped. That entry is not swallowed: post records pass $excluded to Jobs::warning, and Models::options_page does the same for options pages, so both land in bridge/warnings.json and counts.warnings. (Coverage::post_meta only reads post/<id>/… sources, so it ignores these entries; that is fine for a warning.)
• Secrets: the sensitive-field check still runs before renaming, so a secret-named sub-field is dropped under both its key and its name.
• Changelog and version: header, CONTENTRAIN_BRIDGE_VERSION and Stable tag all say 0.6.1. The changelog matches the code: posts and options pages both go through acf_value, and an unowned key is "kept as it is and reported". No tag or release step is in the PR.
• CI: green on this SHA (small, small (10), acceptance, plugin metadata, PHP 7.4/8.3 syntax). I could not run the PHP suite locally (no php here), so I'm relying on CI for that.

Non-blocking, for a later release:

1. A flexible field merges every layout's sub-fields into one by_name map, so a name shared by two layouts with different types resolves to the last layout. Rows keyed by key (the Bridge path this PR fixes) are unambiguous. Choosing the layout from acf_fc_layout would remove the ambiguity for already-named rows.
2. If one row held both a key and its name for the same sub-field, the later cell overwrites the earlier silently. ACF doesn't produce that shape, so I'm only mentioning it.