Upgrade to NumPy 1.22.*

[rajadain]

Overview

This fixes MMW installs because Numba 0.55 is compiled against NumPy 1.22.

This also fixes these messages in MMW:

Unable to import compiled InitSnow_inner, using slower version
Unable to import compiled InitSnowYesterday_inner, using slower version
Unable to import compiled AMC5_yesterday_inner, using slower version
Unable to import compiled CNum_inner, using slower version
Unable to import compiled CNumImper_inner, using slower version
Unable to import compiled CNumPerv_f_inner, using slower version
Unable to import compiled Percolation_inner, using slower version
Unable to import compiled DeepSeep_inner, using slower version
Unable to import compiled AdjUrbanQTotal_inner, using slower version
Unable to import compiled UrbLoadRed_inner, using slower version
Unable to import compiled WashPerv_inner, using slower version
Unable to import compiled WashPerv_inner, using slower version
Unable to import compiled UnsatStor_inner, using slower version
Unable to import compiled DeepSeep_inner, using slower version
Unable to import compiled UnsatStor_inner, using slower version

Closes #100

Testing Instructions

• Pull the latest version of Upgrade to Ubuntu 22.04, Python 3.10 model-my-watershed#3631
• Destroy and recreate the app and worker VMs to pull this version of GWLF-E

  vagrant destroy -f app worker && vagrant up

• Run ./scripts/debugserver.sh
  • Ensure there are no GWLF-E warnings
• In another terminal window, run ./scripts/debugcelery.sh
  • Ensure there are no GWLF-E warnings
• Open http://localhost:8000/ and draw an arbitrary shape
• Run the Watershed Multi Year Model
  • Ensure it completes successfully
  • Ensure there are no warnings in the Celery output

[rachelekm]

Success! I'm able to successfully complete the actions in the testing instructions and have no no GWLF-E warnings 👍

[rajadain]

Thanks for reviewing!