[lestarch-autobot]

Binds FPP hook queue-full behavior to Python. When a component's queue is full, F Prime calls an overflow hook instead of queuing the message; this PR generates the C++ override that forwards each hook into Python plus a Python template stub.

Member	Generated hook	Arguments
async input port p: P hook	p_overflowHook	portNum + port args
internal port i(...) hook	i_overflowHook	internal port args
async command C(...) hook	C_cmdOverflowHook	opCode, cmdSeq only

Changes

• view.py: OverflowHookView, a synthetic HandlerView for each hooked member, appended to ComponentView.handlers, so the existing generators emit the forwarding override and stub unchanged. The old "hook unsupported" rejection is removed.
• component_generator.py: the generated init throws pybind11::attribute_error at startup if the Python class lacks any overflow hook (hooks only run on overflow, so a missing one would otherwise surface on the caller's thread at the worst moment). The lookup is on the class (so the base class's fall-through to the C++ object cannot satisfy it) and m_self is released before throwing. The command-hook template stub answers BUSY, since F Prime drops an overflowed command unanswered.
• README.md: new "Queue-Full Overflow Hooks" section (naming, arguments, caller-thread/GIL cost, concurrency, buffer ownership, command response, exceptions); hook removed from the unsupported-features table.
• Tests: tests/fixtures/Hooks.fpp + command stand-ins in tests/fixtures/Config.fpp; generator tests for hook signatures, GIL acquisition, Python stubs, the init check, and no hook for drop/unhooked members.
• SDD: no change. The repo's only SDD (FprimePython/PythonRateGroupDriver/docs/sdd.md) covers the rate-group driver, not the autocoder; the README is the autocoder's design doc.

Not in scope

• An exception raised inside a Python hook is not caught, the same as every other forwarded handler; a common exception boundary remains future work.

Validation

• mypy (strict) clean; pytest tests 13/13.
• fprime-python-reference with a temporary hook on PythonGc.referenceTracked and a hooked internal port: builds; integration tests 4/4; flooding a 10-deep queue with 64 internal messages called the Python hook 54 times.
• Same deployment with referenceTracked_overflowHook removed from the Python class: setup raises PythonGc does not implement overflow hook referenceTracked_overflowHook. The reference fsw_main.py then calls teardown on the half-built topology, which asserts in BufferManager::cleanup (m_buffers != nullptr); that happens after any setup failure and is outside this PR.
• Local multi-lens review (code, correctness, design, architecture, operational, docs, maintainability, tests, security, supply chain), nothing posted to GitHub; all must-fix findings addressed.

Co-authored-by: M Starch LeStarch@googlemail.com