Enable deploying chrome-tests-syncer independently

[ViniciustCosta]

Motivation

Decouple the chrome-tests-syncer code releases from the rest of clusterfuzz, in order to enable it to be deployed independently of the clusterfuzz instances by the chrome fuzzing team .

Changes

• Extended the butler deploy script to produce a dedicated deployment package for "--release=chrome-tests-syncer" and to produce both the prod and tests-syncer packages when releasing to prod.
• Changed chrome-tests-syncer Dockerfile/start.sh to pull from this dedicated deployment package.
• Moved and edited the tests-syncer code to be executed as a k8s cronjob.
• (side effect) Updated unit test to account for deploying the chrome-tests-syncer manifest.

Tests

• Deployed a chrome-tests-syncer release using this PR and produced the dedicated artifact (https://screenshot.googleplex.com/8W88DmYwFKPkSfZ.png).

• Built an updated chrome-tests-syncer image that pulls from this artifact (https://screenshot.googleplex.com/8HqEvV7bnBggBPj.png).

• Created the chrome-tests-syncer cronjob (along with PR in clusterfuzz-config: 2428107) pulling from this updated image, applied it to the GKE and it ran successfully:
  

• Mocked a deployment with --release=prod (by printing the command instead of actually uploading to gcs):

Bug: b/389048599

[vitorguidi]

• Extended the butler deploy script to produce a dedicated deployment package for "--release=chrome-tests-syncer" and to produce both the prod and tests-syncer packages when releasing to prod.
• Changed chrome-tests-syncer Dockerfile/start.sh to pull from this dedicated deployment package.
• Moved and edited the tests-syncer code to be executed as a k8s cronjob.

Please describe the testing strategy adopted, and post evidence that the change does what it is meant to do (for instance, the manual deployment for isolated test syncer that we performed)

[jonathanmetzman]

LGTM with nit

[vitorguidi]

lgtm