Fix - remove BIG Sync handles on BigSync.terminate to prevent resync failures

[markusjellitsch]

The Issue

BigSync.terminate() sent the terminate command and set the sync
state, but never removed the sync's handle from the device's big_syncs dict nor its BIS link
handles from bis_links. Only the on_big_sync_lost callback did that cleanup.
Because the stale entries were never removed,
re-syncing the same BIG repeatedly leaked entries until the 240th re-sync (0x00-0xF0 valid range) failed.

How to reproduce

The following test with transmitter (creating a big) and receiver (syncing to the big, then terminating) demonstrates the issue:

 File "/home/maxl/dev-projects/rootcanal/examples/python/broadcast_stress_test.py", line 406, in main
    raise RuntimeError("Stress test failures:\n" + "\n".join(failures))
RuntimeError: Stress test failures:
session 240 (unencrypted): RuntimeError("session 240 failed during create BIG sync: OutOfResourcesError('All valid BIG handles already in use')")

The Fix

This PR fixes the above described issue by adding a helper method _remove_big_sync that pops the big/bis handles once the BIG sync was terminated. this method is also used in on_big_sync_lost().
In addition , test_big_sync_and_terminate_255_times was added,, which creates and terminates a BIG sync 255
times and asserts no stale entries accumulate across the full handle wrap-around.

[zxzxwu]

Thanks for the fix! Left a couple of small comments.