…a leaf-list (#4681)

#### Why I did it

The GCU (`generic_config_updater`) patch sorter aborted with an unhandled `ValueError ("'<x>' is not in list")` when a `config replace` / apply-patch changed a create-only PORT field (e.g. `lanes` during a breakout) while that port was a member of a multi-member leaf-list such as `ACL_TABLE.ports`.

During move validation the sorter transiently removes the port from the leaf-list in the simulated intermediate config, but a still-present leafref to it then fails to resolve (`list.index` raises `ValueError`). The exception propagated out of `RemoveCreateOnlyDependencyMoveValidator._validate_member` and aborted the whole sort, so any GCU/config-replace-based breakout failed on a realistic switch config (auto-rollback). The native `config interface breakout` path avoids GCU and is unaffected.

#### How I did it

Treat an unresolvable reference in a simulated intermediate config as an *invalid move*: catch the error in `_validate_member` and return `False` so the DFS backtracks to a valid ordering instead of letting the exception propagate and abort the whole sort.

#### How to verify it

Added a regression test that reproduces the create-only PORT field change against a multi-member `ACL_TABLE.ports` leaf-list and asserts the move is rejected rather than raising:

```
python3 -m pytest tests/generic_config_updater/patch_sorter_test.py \
 -k test_validate__unresolvable_ref_in_simulated_config__move_rejected
```

Verified the new test **fails** with the original code (raising the reported `ValueError`) and **passes** with the fix.

#### Which release branch to port

- [x] 202605

Signed-off-by: Sonic Build Admin <sonicbld@microsoft.com>