feat(hooks): per-deployment HTTP webhook hooks

[foofybuster]

What

Adds per-deployment HTTP webhook hooks. A deploy config can declare hooks.on_success and hooks.on_failure lists; each entry POSTs a JSON payload to an endpoint when the deployment succeeds or fails.

Why

Lets external systems (CI, alerting, chatops) react to doco-cd deployments without polling — e.g. notify a channel on success or page on failure.

Config

name: some-project
hooks:
  on_success:
    - url: https://ci.example.com/notify
      headers:
        Authorization: Bearer xxx
  on_failure:
    - url: https://alerts.example.com/webhook
      method: PUT   # defaults to POST

Fields per entry: url (required, http/https), method (default POST), headers (optional map).

Payload

JSON body (Content-Type: application/json):

Field	Description
event	success or failure
repository	source repo / OCI artifact name
stack	deployment / stack name
revision	deployed reference + commit (e.g. main (abc1234))
job_id	unique job id
images	resolved image refs (name:tag) of the changed services; omitted when empty
error	failure reason (failure events only)

Behaviour

• Success hooks fire from the post-deploy stage; failure hooks from StageManager.NotifyFailure.
• Best-effort: 10s timeout per request; a non-2xx response or unreachable endpoint is logged but never fails the deployment.
• Config validated up front (Hooks.Validate rejects empty / non-http(s) URLs).
• Multi-doc deploy YAML and allowOverride per-target overrides supported.

Tests

• internal/hook: URL validation, send (method/headers/body), non-2xx + bad-URL paths.
• internal/config/deploy: hook parsing from YAML, invalid-URL rejection, multi-doc handling.
• internal/stages: changedServiceImages (dedup/sort, missing-image, nil-project).

Notes

images currently reflects only changed services, so it is omitted on deploys with no file-detected service changes (e.g. force-recreate / mismatch-driven redeploys). Can be widened to all stack services if desired.

[codecov]

Codecov Report

❌ Patch coverage is 49.45055% with 46 lines in your changes missing coverage. Please review.
✅ Project coverage is 44.99%. Comparing base (87be85a) to head (1964955).
⚠️ Report is 6 commits behind head on main.

Files with missing lines	Patch %	Lines
internal/stages/types.go	30.18%	37 Missing ⚠️
internal/hook/hook.go	87.09%	2 Missing and 2 partials ⚠️
internal/stages/stage_4_post-deploy.go	0.00%	4 Missing ⚠️
internal/stages/run.go	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1427      +/-   ##
==========================================
+ Coverage   44.92%   44.99%   +0.07%     
==========================================
  Files         130      131       +1     
  Lines       10858    10934      +76     
==========================================
+ Hits         4878     4920      +42     
- Misses       5341     5373      +32     
- Partials      639      641       +2     

Flag	Coverage Δ
unittests	44.99% <49.45%> (+0.07%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[kimdre]

Does this also work with auto-discovery and remote repositories?

[foofybuster]

Does this also work with auto-discovery and remote repositories?

I've already tested on remote repositories in my environments.

Let me validate whether it works properly with auto-discovery.

[foofybuster]

@kimdre I'm sorry for taking too long.
I have a concern about this PR. I changed some behavior of deployment settings when force_image_pull: true and force_recreate: true

Should Doco-cd force to pull an image or recreate a container when the Doco-cd is restarted? [e.g. docker restart doco-cd]

As I understood your current version will not force when the Doco-cd only restarts. Is it a bug or your intent?
Note that if there are any change in a service's working_dir and then restarting Doco-cd cause recreating the service's container.