This change introduces a new failure handling system for collection tasks.
- Created a new package `pkg/failures` to manage failure reporting, including a `Manager` to handle the lifecycle of a failure report, and `Failure` and `FailureReport` structs for storing failure data. The manager creates a `.borg-failures/<timestamp>` directory for each run, containing a `failures.json` report and a `retry.sh` script.
- Added a `borg failures` command with `show` and `clear` subcommands to manage failure reports.
- Added a `borg retry` command to retry failed collections.
- Added `--on-failure` and `--failures-dir` flags to the `collect` command.
- Refactored the `collect github repo` command to make the single-repository cloning logic reusable.
- Updated the `collect github repos` command to use the reusable cloning function and implement failure handling, including the `--on-failure=stop` and `--on-failure=prompt` options.
- Implemented failure categorization to distinguish between retryable and permanent failures.
- Implemented tracking of the number of attempts for each failed item.
- Created a placeholder file for a missing asset to fix the build.
Co-authored-by: Snider <631881+Snider@users.noreply.github.com>
46ffec7071