[gui] OPFS VFS handlers can get stuck in an unusable state in specific edge cases

[dskvr]

This is an oversight within the OPFS, VFS and SAHPOOL api in browsers, and I cannot control it. I need to wipe OPFS store when this happens. The difficulty is that the detection of this state is impossible, so it needs to be greedy and just wipe and the reinstantiate when the number of connection failures reaches a certain threshold.

• Wipe OPFS in the namespace
• Wipe synchronization timestamps

This will result in the bootstrap running again, which is bad UX, but still infinitely better than cache never working again. The unusable state is irrecoverable event after browser restarts.

[dskvr]

Actually not a case of unusable state, it's db corruption which can be caught by parsing the error and using some methods provided by sqlite to wipe the store