feat: improve performance of the "Saving deployment state" step for Node projects

[ErlendHer]

Description of changes

When uploading studio backend files, there is a glob that scans through all files/folders in the amplify project directory. For Amplify projects where there are multiple Node projects, this is problematic as scanning through node_modules can be very resource intensive.

Below are some benchmarks that were performed for my teams Amplify project (we currently have 40 Node lambda functions, and 6 custom resources written in JavaScript and thus having node_modules)

This is the benchmark before the suggested change below

Task	Time
Zipping the #current-cloud-backend	1.886s
Uploading the .zip file	7.487s
Deleting the old S3 directory from studio-backend	898ms
Performing the glob on the amplify directory	6 minutes 41 seconds
Uploading the studio-backend files to S3	558ms

After running the same benchmark with the proposed changes, all of the steps remained roughly the same, except for
the glob which was reduced to 1.3 seconds. For our project, this is very significant, as this saves us almost 7 minutes per deploy.

Issue #13301

Description of how you validated changes

Ran amplify push both before and after and benchmarked the results for our production project. Noticed closed to a 7 minute decrease in deployment time

Checklist

• PR description included
• yarn test passes
• Pull request labels are added

All tests are passing except for the featureFlags.test.ts, which should not be relevant to the changes I've made, not sure if this is a local setup issue or if tests in dev are not passing atm

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.