[gmmcosta15]

Description

• Feature
• Bug Fix
• Code Refactor
• Documentation

Fixes two Happy Hare and connection problems seen on AMU printers: FlowGuard clogs being reported as tangles, and the screen getting stuck on "Klipper down" after Klipper had recovered.

Changes

• FlowGuard rule split by the type Happy Hare reports, which is the same word Mainsail's Clog/Tangle Detection meter shows:
  • FlowGuard detected a clog / A clog has been detected on ... -> "Filament Clog Detected"
  • FlowGuard detected a tangle / A tangle has been detected on ... -> "Filament Tangle Detected"
  • A clog/tangle has been detected (encoder setups, HH can't tell which) -> "Filament Clog or Tangle"
• moonrakerComm.on_error: now only logs; removed the unused self.disconnected

Motivation

• Release-2.0 matched every message containing flowguard to "Filament Tangle Detected", so a real clog showed as a tangle. dev only matched the tangle text, so a clog had no specific message.
• websocket-client (1.9.0) passes exceptions raised inside our callbacks to on_error without closing the socket and without calling on_close. The old on_error set connected=False, after which every send_request, including the server.info poll, was dropped and reconnect() never retried because the ws thread was still alive. The "shutdown" notification kept arriving but the "ready" that clears it never came, so the screen stayed on "Klipper down" until BlocksScreen was restarted (seen on an RF50 running 8408a6e on 10-13, 10-14 and 10-15). A real drop always ends in teardown() -> on_close, which already sets connected=False; KlipperScreen's on_error also only logs.

Tests

• New unit tests:
  • FlowGuard clog, tangle and clog/tangle messages
  • a raising callback leaves the socket open and usable

[RobeMartins]

happy hare m117 filter is not needed since happy hare has a flag to disable them