Alcove adapter: detect() rejects sessions with 'id' instead of 'session_id'

[decko]

Problem

The Alcove adapter's detect() method only matches files containing the string "session_id" in the first 4KB:

def detect(self, source: Path) -> bool:
    ...
    header = file_handle.read(DETECT_READ_SIZE)
    return '"session_id"' in header and '"transcript"' in header

The Alcove Bridge API returns sessions with "id" as the top-level identifier, not "session_id". When exporting sessions via alcove status <id> --full, the output looks like:

{
  "id": "12052e41-...",
  "transcript": [...]
}

The "session_id" string may appear deeper in the transcript (inside the system init entry), but if it's beyond the 4KB detection window, the session is skipped.

Similarly, load() accesses raw["session_id"] which raises KeyError for files that only have "id".

Impact

In testing with 92 exported Alcove sessions, 37 were skipped (40%) due to this detection mismatch.

Suggested fix

detect(): Accept either field:

return ('"session_id"' in header or '"id"' in header) and '"transcript"' in header

load(): Fall back from session_id to id:

session_id = raw.get("session_id") or raw["id"]

Related

This is separate from #176 (hardcoded rework_cycles/findings). That issue is about what happens after loading; this one is about sessions never being loaded at all.

[decko]

Closing — 0.9.0 already handles id fallback. The actual skip cause is empty transcripts (transcript: []) where the field doesn't appear in the first 4KB detection window. Will file a separate issue if needed.