Give the Influx and MQTT nodes their two missing knobs

The Influx client was built with no timeout, so every query and write fell
through to influxdb-client's own 10 s default — invisible to a flow and
unchangeable. The param is in seconds like its peers; the client counts in
milliseconds, so the call sites convert.

The publisher backlog was a module constant, read once at import. It is the
depth at which the oldest payload is dropped and the node goes degraded, and
a node that bursts wants more than one that trickles, so it moves to Params
and is read where the queue is built.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K1moruzue2kTJd3uVisgNk

backend/fluksio/flow/nodes/mqtt.py

@@ -19,10 +19,6 @@ if TYPE_CHECKING:
 
 logger = logging.getLogger(__name__)
 
-# Deep enough to ride out a broker hiccup, shallow enough that a publisher
-# which cannot keep up drops old values instead of growing without bound.
-PUBLISH_QUEUE_SIZE = 256
-
 
 def topic_matches(filter_: str, topic: str) -> bool:
     """Does an MQTT topic filter cover this topic?
@@ -85,6 +81,7 @@ class MqttNode(Node):
         - ``keepalive`` (int): Keepalive interval in seconds (default: 60)
         - ``timeout`` (float): Deadline for a broker operation in seconds
           (default: 10.0)
+        - ``publish_queue_size`` (int): Publisher backlog depth (default: 256)
     :type params: dict
     :param name: Optional name for the node.
     :type name: str | None
@@ -154,6 +151,7 @@ class MqttNode(Node):
         "retain",
         "keepalive",
         "timeout",
+        "publish_queue_size",
         "json_keys",
         "_topic_to_ports",
         "_wildcards",
@@ -187,6 +185,17 @@ class MqttNode(Node):
             gt=0,
             description="Give up on a broker operation after this many seconds.",
         )
+        # Deep enough to ride out a broker hiccup, shallow enough that a
+        # publisher which cannot keep up drops old values instead of growing
+        # without bound. A node that bursts wants more than one that trickles.
+        publish_queue_size: int = Field(
+            default=256,
+            gt=0,
+            description=(
+                "How many payloads may wait for the broker. Past this the oldest "
+                "is dropped and the node reports degraded."
+            ),
+        )
         # Which key to lift out of a JSON object payload. A device that wraps
         # its reading — Victron's ``{"value": 5}`` — is otherwise a Python node
         # per port. One key for every port, or a per-port mapping.
@@ -260,6 +269,7 @@ class MqttNode(Node):
         self.retain = cfg.retain
         self.keepalive = cfg.keepalive
         self.timeout = cfg.timeout
+        self.publish_queue_size = cfg.publish_queue_size
 
         # Runtime state
         self._subscription_task: asyncio.Task[None] | None = None
@@ -467,7 +477,7 @@ class MqttNode(Node):
         """Run the task that owns this node's connection to the broker."""
         if self._publish_queue is not None:
             return
-        self._publish_queue = asyncio.Queue(maxsize=PUBLISH_QUEUE_SIZE)
+        self._publish_queue = asyncio.Queue(maxsize=self.publish_queue_size)
         self._loop = asyncio.get_running_loop()
         self._publisher_task = self._run_supervised("mqtt-out", self._publisher_loop)