[feature]: Add diagnostics mode to capture proof failures and support artefacts

[ffranr]

The goal is to introduce an extendable diagnostics feature that can aggregate troubleshooting data across tapd subsystems. A user should be able to zip the entire diagnostics directory and provide it to support or link it in a GitHub issue. In the first iteration we will focus on troubleshooting proof-validation failures during asset transfer.

Scope for first iteration

• Add --diagnostics-dir (and config field) to enable diagnostics. Diagnostics mode is disabled unless dir path is given.
• When in diagnostics mode:
  • On tapd start, create a new run directory under the diagnostics directory. Naming: ts<unix_ts>-pid<pid>.
  • After anchor tx broadcast during asset transfer, if generated output proofs fail validation, store those proof artefacts in the active diagnostics run directory for support.

Architecture notes

• Implement diagnostics as its own sub-service inside tapd that manages filesystem layout under the diagnostics directory.
• Other subsystems can hand artefacts to the diagnostics service, which decides placement and storage.
• Diagnostics runs non-blocking to avoid impacting normal tapd operation.

Future directions

• Dump tapd version and build metadata in the run directory.
• Optional trace logging directed to the run directory.
• diagnostics handles goroutine dumps and pprof capture under certain conditions.