Update deployment workflow to allow manual deployments from tag

Currently the deployment workflow in WATcher is different from CATcher.

CATcher:
1. PR to main, updating the version
2. Create version tag for the commit we want to deploy (e.g. v3.6.0)
3. Manually trigger the deployment workflow from the commit

WATcher:
1. PR to main, updating the version
2. PR from main to `deploy`
3. Deployment runs automatically when there are changes to the `deploy` branch

We think that the CATcher workflow provides more flexibility (and is less troublesome as we only have to PR once, whereas in WATcher we have to PR twice), so we are requesting to change the workflow in WATcher to follow the workflow in CATcher.

Start by locating WATcher's deployment workflow and compare its trigger and branch behavior with the CATcher workflow described in the issue. Update the workflow so a deployment can be manually triggered for a selected version tag, then verify that the resulting process needs only the version-update PR and tag.

Written by the indexing model from the issue text.