fix: add start_time to ProcessUID on StorageLock

[joshieDo]

Makes the identifier more unique by adding start_time.

Fixes the following kind of issue:

1. reth node with PID 9 locks storage.
2. reth node stops non-gracefully leaving the lockfiles behind.
3. firefox starts with PID 9.
4. reth node attempts to start but realizes that PID 9 is executing, and thus, does not acquire the lock.

[mattsse]

I think this should work

[joshieDo]

Calling sysinfo::System to get the process start_time is too expensive during ef-tests. So, added the feature disable_lock, which will make StorageLock::try_acquire always succeed, and use it during ef-tests.

[mattsse]

pedantic suggestions

[mattsse]

if this feature enabled do we then need sysinfo dep?

opting out is a lot harder than opting in, but in this case we expect that this is only disabled for testing, so imo this is fine