Orchestrator: Proper handling for SAVE-cli issues reported by SAVE-agent

[orchestr7]

trafficstars

1. Right now if the processing fails we simply mark agent execution as CLI_FAILED
2. In this case orchestrator simply stops ALL agents (if all of them are IDLE or CLI_FAILED)
3. But tests are not marked as FAILED - they are still in RUNNING state

We need to do the following:

• Rename CLI_FALED to INTERNAL_SAVE_CLI_ERROR
• Mark tests FAILED (fail all of them)
• Think of how to give extra information to the user of what happened and why Agents failed

[orchestr7]

There is already some work have been done in https://github.com/analysis-dev/save-cloud/pull/776

In this PR we fail execution. But we need to get the reason of failure and provide it to the user

[orchestr7]

Description of issue a little bit different, but topic is fine: https://github.com/saveourtool/save-cloud/pull/2635/files#r1341720749