chore: Hook up configuration options for BadTokenMonitor

[MartinquaXD]

Background

We should be able to configure the driver side bad token detection for each solver individually since they might support different tokens.

Things that need to be configurable on a per-solver basis:

• list of explicitly supported tokens
• list of explicitly unsupported tokens
• how long a token should be marked as unsupported (to err on the safe side tokens should probably not be banned forever immediately)
• parameters needed to configure the heuristic
  • depends on the actual heuristic and implementation - could be sth like:
    • minimum of observed failures before deny-listing may happen
    • ratio of working vs. revering solutions before deny-listing
• which mode to run (enabled vs. logOnly)

To make things easier all these should have reasonable defaults so the feature can start running without having to merge a huge infra PR before.