Provider uptime checks caching bad connection

[Redm4x]

If a provider goes down the uptime checks in the indexer app will flag it as offline. When it comes back up it should be flagged as online, but in some cases the checks will continue to fail even when the provider is back up. Restarting the indexer fixes the issue so it points to a caching issue. The grpc clients are memoized so that could be related. grpc-js also seems to reuse existing connection even when recreating the client.

Here's 2 provider that had this issue:

• provider.mylibra.org | akash18gewxdlc5llzgdr75x3ls4w4cte8hgfzsnr27s
  • Happened on 2024-11-17 & 2025-01-04,
• provider.akash-palmito.org | akash15ksejj7g4su7ljufsg0a8eglvkje94z8qsh68a
  • Happened on 2024-12-02 (discord thread)
• provider.dcnorse2.ddns.net
  • Happened on 2025-01-23 (discord threadf)

This ticket is for documenting future occurrences of the bug and ideally try to reproduce it so that we can figure out the root cause and fix it.