Facade error handling

[staszek-krotki]

Make error handling consistent and resilient so that we are able to read logs and trace back to the original issue.

Problems:

• How to provide the most informative exception message when Provider/Yagna fails to start?
• How to handle errors (after start) from ya-runtime-ai in informative way?
• Make exceptions more meaningful, reading logs should be last resort

Proposals:

• Failing activity sets error message, we could expose it in exception

Testing for scenarios:

• yagna or provider fails silently (example: file-watch crashed the app)
• api errors other than unauthorized
• cmd error (e.g. when presets are modified but others as well)

For testing purposes try to modify yagna/provider to trigger different errors (fails, crashes, deadlocks)

The end goal is to expose these exceptions as event as part of this task: https://github.com/golemfactory/gamerhash-facade/issues/104

[staszek-krotki]

Sample Api error:

[staszek-krotki]

Provider startup error:

[staszek-krotki]

Normal Stop (no error):

[staszek-krotki]

Yagna CLI errors (e.g. yagna payment --init) are visible in logs but no exception/error code returned. Errors can be seen later:

[staszek-krotki]

After changes in CLI handling: