sync-v2 step 6: memory + connection bounds (§7, §4)

- concurrency.loop_semaphore: per-event-loop semaphores (safe across pytest loops
  and process restarts via a WeakKeyDictionary keyed by the running loop).
- provider: holds a per-account (≤4) and per-host (PER_HOST_CONN_CAP) slot for the
  whole lifetime of an IMAP socket, released on close; SMTP send takes a per-host
  slot for its duration. A worker can no longer open hundreds of sockets to one
  host/account.
- indexer: parsing goes through a PARSE_CONCURRENCY semaphore, and a message over
  STREAM_PARSE_MB is parsed from its blob file (parse_message_file) instead of an
  in-memory tree, so peak RAM ~= PARSE_CONCURRENCY x STREAM_PARSE_MB regardless of
  mailbox sizes. Backfill parallelism stays bounded by the bulk queue only.
- config: PER_HOST_CONN_CAP, PARSE_CONCURRENCY, STREAM_PARSE_MB.
- tests: per-account cap blocks the 5th connection / distinct accounts independent;
  file-vs-bytes parse equivalence + large-message streaming routing.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

backend/tests/indexer/test_streaming_parse.py

@@ -0,0 +1,60 @@
+"""Bounded / streaming MIME parse (refactor §7).
+
+Large messages are parsed from their blob file rather than an in-memory tree, and
+all parsing passes through the PARSE_CONCURRENCY semaphore.
+"""
+
+from __future__ import annotations
+
+import uuid
+from email.message import EmailMessage
+
+from app.indexer import indexer
+from app.indexer.mime import parse_message, parse_message_file
+
+
+def _raw(body: str = "hello") -> bytes:
+    msg = EmailMessage()
+    msg["Subject"] = "Streamed"
+    msg["From"] = "a@example.com"
+    msg["To"] = "b@example.com"
+    msg["Message-ID"] = "<s1@example.com>"
+    msg.set_content(body)
+    msg.add_attachment(
+        b"x" * 1024, maintype="application", subtype="octet-stream", filename="a.bin"
+    )
+    return msg.as_bytes()
+
+
+def test_parse_file_matches_bytes(tmp_path) -> None:  # type: ignore[no-untyped-def]
+    raw = _raw()
+    path = tmp_path / "msg.eml"
+    path.write_bytes(raw)
+    # Parsing from the file handle yields exactly the same ParsedEmail as from bytes.
+    assert parse_message_file(path) == parse_message(raw)
+
+
+async def test_parse_bounded_streams_large_messages(monkeypatch) -> None:  # type: ignore[no-untyped-def]
+    from app.config import get_settings
+
+    threshold = get_settings().stream_parse_mb * 1024 * 1024
+    calls: list[str] = []
+
+    def fake_file(path: object) -> str:
+        calls.append("file")
+        return "file"
+
+    def fake_bytes(raw: bytes) -> str:
+        calls.append("bytes")
+        return "bytes"
+
+    monkeypatch.setattr(indexer, "parse_message_file", fake_file)
+    monkeypatch.setattr(indexer, "parse_message", fake_bytes)
+
+    user_id = uuid.uuid4()
+    small = await indexer._parse_bounded(user_id, "sha", b"x" * 16)
+    big = await indexer._parse_bounded(user_id, "sha", b"x" * (threshold + 1))
+
+    assert small == "bytes"  # small stays in memory
+    assert big == "file"  # over threshold => parsed from the blob file
+    assert calls == ["bytes", "file"]