A list of essential error handling

Notes: for ADACS development

1. For any major error, peacefully skip this dataset/beam without interupt routine pipeline for other/future datasets. Save an error log file detailed the reason.
2. Save the output info for each run.

Potential situations

• no available short images - skip this dataset
• not enough short images (lower threshold?) [from dev]
• lack of deep catalogue / unreadable deep catalogue format - skip this dataset
• cannot generate local rms for some reason - skip this dataset
• cannot generate middle products chisquare map, and/or peak map, and/or std map, and/or Gaussian map for some reason
  • lack of std map - skip this dataset
  • lack of partial other maps - use the existed one or two maps to generate final products (e.g., chisq + peak, or chisq + Gaussian, or chisq only)
  • lack of all of other maps (i.e. no chisquare map + no peak map + no Gaussian map) - skip this dataset
• unreadable middle products (chisquare map, peak map, std map, Gaussian map) - same logic as above
  • Candidates() class only need one of chisquare, peak and gaussian map (no need to load all at once - only std map is necessary)
• map contains null value (tests data available)
• lack of deep image / unreadable deep image - skip making deep cutout png (can generate csv, gif without problem)
• no candidates found for a beam - skip output [from dev]

Other improvements

• use some proper arguments (with proper help documentation) for python run_all.py
• process all 36 beams in parallel [from dev]
• make a tar file for final output
• remove fits output [from dev]
• rewrite the output files if they exist, or bring overwrite keyword to the app interface [from dev]

Optional feature list