Deepcell upload loop cleaning

[camisowers]

If you haven't already, please read through our contributing guidelines before opening your PR

What is the purpose of this PR?

Closes #628. Changes how we attempt to re-upload failed fov batches to the deepcell server. Also refactors create_deepcell_output() with cleanlier helper functions. The code will now also check for previously processed batches, to avoid re-zipping and re-uploading already segmented batches to deepcell.

How did you implement your changes

Currently, we attempt to re-upload a failed fov batches until success or 5 retry attempts are reached. A cleaner, more long term fix is to attempt to process each batch once, check if any batch .zip files are missing, and re-run the failed batches. The re-running continues until either a successful output or the function times out.

• Re-run friendly additions. For each batch of fovs:
  1. Check if fovs_batch_1.zip exists, if not generate the input zip file
  2. Check if deepcell_output_fovs_batch_1.zip exists, if not upload input file to deepcell and extract segmentation tiffs

Remaining issues

• [x] Set a time limit for reprocessing attempts.
• [x] Tweak the code so it does not re-upload processed fov batches when the cell is re-run

[camisowers]

Jus a couple of small things.

For the Retry call at line 245 num_retries is not defined.

This PR was actually meant to scrap the retry a certain amount of times logic and instead implement recall run_deepcell_direct() in a loop until success or timeout. Looks like I missed some code that needed to be removed.