Fix IndexError crash in TCP packet parser (packet types 115 and 131)

[imwithsam]

Summary

• The packet_type == 115 and packet_type == 131 handlers in _read_tcp_messages each have several self.home_devices[home_id][idx] lookups with no bounds check.
• The packet_type == 67 handler right below them already guards its equivalent lookup with if int(packet[3]) < len(self.home_devices[home_id]): before indexing — this PR brings the other two handlers in line with that existing pattern.
• Without the guard, a packet whose device index isn't present in the locally cached home_devices list throws an uncaught IndexError: list index out of range, which gets caught by the outer try/except and logged as Error parsing an incoming Cync TCP packet — the packet is silently dropped with no further handling.

Observed this live on a real setup: the "state and brightness change" sub-handler (the one at packet[21]) hit this exact IndexError repeatedly on a bulb whose index wasn't (yet) in home_devices[home_id] for that home. Confirmed the same unguarded pattern exists in 5 other spots across both handlers, so fixed all of them for consistency rather than just the one that happened to crash.

Test plan

• python3 -m py_compile passes on the modified file
• Applied the same fix to a live Home Assistant instance running this integration and confirmed the container starts cleanly with no new errors
• Would appreciate a maintainer/community check on an account that regularly reproduces this crash, to confirm packets that previously crashed the parser are now cleanly skipped instead