Refactor Cloudfront cache invalidation logic

[mdial89f]

Currently, on deployment, the entire ui cloudfront distribution is invalidated. So we upload our static archive to s3, then invalidate the whole cache.

This should be refactored; the suggested way is:

Create a lambda to do individual object invalidations. Set that lambda to trigger on each new or updated file in the s3 bucket via an s3 event. That lambda invalidates that object. So only the objects that need to be invalidated are. And, the pipeline can quit concerning itself with invalidations.

AC:

• The cache invalidation is handled by S3 events + lambda, making it that only the objects that need invalidation on each deploy are invalidated
• The current invalidation logic run by the CI system is removed