Add Backoff and Retry Mechanism for Image Pull

[honeyankit]

This pull request introduces a retry mechanism for fetching Docker images, which includes changes to the ImageService and corresponding test files. The main goal is to improve the robustness of image fetching by implementing retries with exponential backoff when Dependabot hit the rate limit (429 error) while downloading docker images from GitHub package registry.

This PR is created to address the recent spikes in the rate limit issues.

Enhancements to image fetching:

• src/image-service.ts: Added constants MAX_RETRIES and INITIAL_DELAY_MS, and a sleep function to facilitate retry logic. Modified fetchImage to fetchImageWithRetry, incorporating retry attempts with exponential backoff and handling of specific error messages such as "429 Too Many Requests". [1] [2]

Updates to test files to use the new retry mechanism:

• __tests__/cleanup-integration.test.ts: Updated the beforeEach hook to use fetchImageWithRetry instead of fetchImage.
• __tests__/container-service.test.ts: Modified the beforeAll hook to use fetchImageWithRetry for fetching the 'alpine' image.

[honeyankit]

The Docker API typically returns a 429 Too Many Requests error for abuse rate limits. Just to be sure, I added a condition to check for either 429 or Too Many Requests. This way, even if one of these terms is missing, the error can still be detected, allowing for a retry.

[honeyankit]

Steps to make changes to this PR in Codespace:

1. Make changes to the PR
2. Execute the below command in the Codespace terminal after any change to the PR

2.1

npm run lint-check
npm run format-check -- --write
npm run test

2.2

nvm install;nvm use;npm clean-install;npm ci;npm run package

Note: If you do execute step 2.2 and commit the code then CI will fail with the below error:

Run script/check-diff
Detected uncommitted changes after build:
diff --git a/dist/main/index.js b/dist/main/index.js
index c09ccea..8f50b37 1006[4](https://github.com/github/dependabot-action/actions/runs/7720200685/job/21044694134?pr=1156#step:7:5)4
Binary files a/dist/main/index.js and b/dist/main/index.js differ
diff --git a/dist/main/index.js.map b/dist/main/index.js.map
index cc44481..e[5](https://github.com/github/dependabot-action/actions/runs/7720200685/job/21044694134?pr=1156#step:7:6)0840f 100[6](https://github.com/github/dependabot-action/actions/runs/7720200685/job/21044694134?pr=1156#step:7:7)44
Binary files a/dist/main/index.js.map and b/dist/main/index.js.map differ

3. Commit and push the code changes
4. Once PR is approved simply merge the PR. This will make the PR deploy in the production

[honeyankit]

@abdulapopoola I have added the delay of 2 seconds

[Nishnha]

@honeyankit and I considered adding retries to

dependabot-action/src/image-service.ts

Lines 39 to 45 in 2374bec

	try {
	const image = await docker.getImage(imageName).inspect()
	if (!force) {
	core.info(`Resolved ${imageName} to existing ${image.RepoDigests}`)
	return
	} // else fallthrough to pull
	} catch (e: unknown) {

. We originally thought this code checks whether the image exists on the remote server, but it actually checks for the existence of the image on the local server, so we do not need to add retries.