Build Binaries improvements

[agardnerIT]

The workflow to build binaries currently:

• Is triggered manually
• Builds the binaries and commits them back to this repo in the dist folder
• Only build for Windows and Linux (MacOS intel and MacOS Apple silicon are not included yet due to lack of GitHub public runners)

The binaries only need to be build just prior to each new release.

1. The workflow should remain as a manual trigger
2. Committing the binaries back to this repo has a few drawbacks: a) Large files in the Git history b) Causes new commits and an update to the history so others need to do a git fetch and git pull. This can get messy and is not a good user experience.

PyInstaller and nuitka (an alternative I've investigated) only build for the architecture on which the build command is executed. Therefore it is anticipated that until GitHub offers public runners for all common architectures, we'll need to spin up other VMs to run the build process (outside of GitHub) which will (probably) mean this process is somewhat manual.

Proposed Solution

• Binaries should be built manually
• Binaries should be pushed to an S3 bucket (or similar "off site" storage)
• Binaries should NOT be added to the Git history
• Binaries should be (manually?) downloaded and attached to each release
• Future work: We should investigate a way to automate this build and attach binaries to each release in an automated way.

[agardnerIT]

Windows and Linux binaries are built by the (manually triggered) workflow. The workflow pushes to S3. Mac binary (universal binary) are built locally and pushed to S3. All binaries are manually downloaded from S3, deleted and uploaded to the release.