Revert "fix: prevent channel monitor loss when remote backup fails"

[pwltr]

Reverts #286

After more discussion I am reverting this, persisting locally first without ensuring remote backup succeeds has the same amount of risk or more than the previous logic, in the sense that remote and local state become out of sync. Restoring from backup subsequently can cause loss of funds.

[ovitrif]

Reverts #286

After more discussion I am reverting this, persisting locally first without ensuring remote backup succeeds has the same amount of risk or more than the previous logic, in the sense that remote and local state become out of sync. Restoring from backup subsequently can cause loss of funds.

Makes sense to some extent :)
I was thinking that it still unlocks a use-case, but it's all hypothetical:
what if restoring never happens from local but local backup could help confirm some info about the funds, for example to justify a refund… 🤷🏻

Still, it's all just a wild guess, and I don't really know if it's a real possiblility.

[pwltr]

what if restoring never happens from local but local backup could help confirm some info about the funds, for example to justify a refund… 🤷🏻

IIUC you are suggesting we persist locally always but not in a directory that LDK would actually use, for debugging purposes? It's a nice idea. The issue we need to solve only really exists for new channels (isNew), because the push balance is not secured by LDK safety measures. So we could add extra logic just for new channels to always write the channel monitor locally, even if remote backup fails, even if that opens up that other edge case of having two states out of sync. But better then not having any backup at all.