Refactor and bugfix for resolve lock logic

[amyangfei]

Feature Request

Is your feature request related to a problem? Please describe:

A known bug from #2867, if the resolved-ts of a region is fallen back, the resolved ts update event won't be sent in L762.

https://github.com/pingcap/ticdc/blob/67f994db6dbc9121ed2c881feb88c4d97387dcd8/cdc/kv/region_worker.go#L742-L764

In this scenario, the penalty of a region won't increase and resolve lock won't launch until the region is reconnected.

https://github.com/pingcap/ticdc/blob/67f994db6dbc9121ed2c881feb88c4d97387dcd8/cdc/kv/region_worker.go#L359-L376

Describe the feature you'd like:

• fallback resolvedTs event skips the lock resolving unexpectly #3061
• Refine the code logic of resolveLock function in region worker, adding unit test to it without the kv client environment. This many need some mock code for kv storage and region states manager of a region worker.

[maxshuang]

@amyangfei @overvenus is there any good way to simulate "kill db instance and leave some lock in region" ?

[amyangfei]

such as https://github.com/pingcap/ticdc/blob/42cccaa14a7586632563bbe861e2251acfd65490/tests/resolve_lock/main.go#L179