Add memory per retry | Refactor GPU parameters

[RhettRautsaw]

Changes:

• Refactor SLURM account, partition, qos handling logic for cleaner code
• Increase requested Slurm memory by 1.5x on each task retry.
  • Most tasks that fail are due to out-of-memory errors.
  • This addition will increase memory by 1.5x per retry
• As an alternative to the [task_runtime] level changes. I've included an example of how to use the dynamic partition for gpu and non-gpu tasks instead.

[RhettRautsaw]

Added additional commit to only try increasing memory when OOM exit codes (137 or 253) are encountered. Added a logging line to let users know when memory is being increased on a retry.