Cycles 19-20: drop stale issue/PR cross-references, prune unused yaml extra

- Strip 'see #148' and 'see #080' references in cluster.py / pool.py
  — internal tracker numbers don't age well and don't help readers
  outside the original context. Rewrote the surrounding comments to
  carry the relevant information inline.
- Drop the unused 'yaml' optional-dependency (pyyaml was never
  imported anywhere). If YAML cluster-config parsing lands later it
  can be re-added with the importing code in the same change.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: antoineleclair

pyproject.toml

@@ -35,7 +35,6 @@ Issues = "https://github.com/letsdiscodev/python-dqlite-client/issues"
 
 [project.optional-dependencies]
 dev = ["pytest>=8.0", "pytest-cov>=4.0", "pytest-asyncio>=0.23", "mypy>=1.0", "ruff>=0.4"]
-yaml = ["pyyaml>=6.0"]
 
 [tool.hatch.build.targets.wheel]
 packages = ["src/dqliteclient"]

src/dqliteclient/cluster.py

@@ -171,10 +171,10 @@ async def try_connect() -> DqliteConnection:
             return conn
 
         # Retry only transport-level errors. Leader-change OperationalError
-        # codes are now reclassified into DqliteConnectionError at connect
-        # time (see #148), so we no longer need OperationalError in the
-        # retry set — that avoids amplifying a schema/SQL error into 5 ×
-        # N_nodes RTTs before propagating.
+        # codes are reclassified into DqliteConnectionError inside
+        # DqliteConnection.connect(), so we no longer need OperationalError
+        # in the retry set — that avoids amplifying a schema/SQL error
+        # into 5 × N_nodes RTTs before propagating.
         return await retry_with_backoff(
             try_connect,
             max_attempts=3,

src/dqliteclient/pool.py

@@ -366,8 +366,8 @@ async def close(self) -> None:
 
         # In-use connections are closed by acquire()'s cleanup when they
         # return — the else branch checks _closed and closes instead of
-        # returning to the pool.  Force-closing them here would race with
-        # the acquire context manager and corrupt _size (see #080).
+        # returning to the pool. Force-closing them here would race with
+        # the acquire context manager and corrupt _size.
 
     async def __aenter__(self) -> "ConnectionPool":
         await self.initialize()