"Library Log File" Grows With No Limit (2.0)

[As requested](https://github.com/aws/aws-logging-dotnet/issues/80#issuecomment-900447499), opening this ticket that was previously closed prematurely to re-report this issue.

The issue: if there is a problem preventing the logs from being able to be sent to CloudWatch, the `aws-logger-errors.txt` file will be filled and grow with no bounds, eventually filling the disk.

As `sattpat` replied back in 2019:

>We certainly dont want to be logging every single issue. We would instead like to capture exceptions that are unique. Meaning that if you get exceptions as a result of network connection, we only log that once and we skip the repeats.

This seems like a very good solution. Perhaps whatever is logging to `aws-logger-errors.txt` could keep track of recently written errors, and if it's logging the same error several times in a row, it logs an additional "This issue won't be logged again until the process is restarted", and then stops logging that same error again.

Use the aws-logger-errors.txt writer and the CloudWatch delivery failure path as entry points. Determine how repeated exceptions are currently recorded and define the duplicate-suppression behavior described in the issue. Done means repeated identical failures no longer grow the file without bounds, while unique errors remain available for diagnosis.

Written by the indexing model from the issue text.