If a background task is created, that constantly fails with a retryable error, VFS4G is stuck

If we have a background task, that is constantly failing due to a retryable error, GVFS.Mount process ends up consuming all available CPU and the whole enlistment is frozen. To avoid that, we need to:

1. Better classify which errors are retryable and which are not. Today all IOExceptions are treated as retryable, but those include things like System.IO.PathTooLongException, which is unlikely to succeed after retrying.
2. Introduce a retry limit on background tasks. If we've retried the task for thousands of times, perhaps it's worth to at least move it to the back of the queue, or give up entirely. Or at maybe terminate GVFS.Mount process, emitting problem description to logs, so that it's easy to identify, what the issue was and what's causing it.

Start in the GVFS.Mount background-task retry handling, then trace how IOExceptions such as System.IO.PathTooLongException are classified and how repeatedly failing tasks remain queued. The issue names no files or tests; completion would require an agreed retry policy, bounded handling for persistent failures, and useful problem details in logs.