Suggestions from Discord on reliability

First improvement suggestion:

• When a new introduced settings "reload on failed load" is enabled (default: false, so updating users do not get it enabled by default) and "reload max retries" is not reached, the filament is unloaded similar to the LANE_EJECT macro with the difference that it stops after "untriggering" the hub sensor. This prevents full bowden length unload into nirvana. And rerun the load cmd again
• When the max retries has been reached, an error is thrown

Second improvement suggestion:

• When loading, a new state for "attempt load" is introduced and set when the filament triggers the hub sensor. This setting is "var saved".
• The state is change to the loaded state after the operation was successful. (var saved as usual)
• When something happens to result in a firmware-restart or similar, the BT can do the prep and acknowledges which filament is still in the bowden.
• Calling BT TOOL UNLOAD in a state of "attempt load" unloads the filament in steps like in the LANE_EJECT command without going all out (as described in first suggestion)

Both improvements have the following effects:

1. a failed load has the chance to be self-remediated by unloading it back to in front of the hub sensor, resulting in reliable print finishes without intervention
2. if it errors out or the configuration has this feature disabled, the manual intervention is to hit "BT_TOOL_UNLOAD" to unload the filament instead of trying to fix it by manual parameterized calls of BT_LANE_MOVE

https://discord.com/channels/1229586267671629945/1359941893013897346/1360919767556558858

Start by tracing the BT_TOOL_UNLOAD, BT_LANE_MOVE, and LANE_EJECT entry points named in the issue, then review how load failures and filament state are currently handled. Done means failed loads can retry when enabled, stop unloading at the hub sensor, report an error after the retry limit, and let BT_TOOL_UNLOAD recover an attempted load.