Add per-route forward-health-checks option

[kemko]

Summary

Some HTTP services behind reproxy expose their own /ping or /health endpoints with application-specific responses that clients depend on. For example, audiobookshelf's /ping must return {"success": true} as a JSON object — its clients use this to verify the server is reachable and functioning correctly. Currently, reproxy's built-in pingHandler and healthMiddleware intercept all GET /ping and GET /health requests globally (responding with "pong" and a health JSON summary respectively) before route matching even runs, so these requests never reach the backend. There is no way to disable this per route or per domain.

This PR adds a new per-route option forward-health-checks (supported across all providers: Docker labels, file config, Consul Catalog tags, and static rules) that forwards /ping and /health requests to the backend instead of reproxy handling them. The option is off by default, so existing behavior is fully preserved.

Docker example

audiobookshelf:
  image: ghcr.io/advplyr/audiobookshelf
  labels:
    - "reproxy.server=audiobookshelf.example.com"
    - "reproxy.route=^/(.*)"
    - "reproxy.dest=/$1"
    - "reproxy.forward-health-checks"

File provider example

audiobookshelf.example.com:
  - { route: "^/(.*)", dest: "http://192.168.1.10:8080/$1", forward-health-checks: true }

Test plan

• All existing tests pass (go test -race -timeout=60s -count 1 ./...)
• New tests for Docker provider with forward-health-checks label
• New tests for Consul Catalog provider with forward-health-checks tag
• Updated file provider tests with forward-health-checks YAML field
• New static provider tests for 5th element parsing
• New proxy tests: /ping and /health forwarded when flag is set, built-in response preserved when not

[kemko]

Is there anything I can do to prevent this PR from being considered junk? Maybe I should file an issue?

[umputun]

Is there anything I can do to prevent this PR from being considered junk? Maybe I should file an issue?

Hmm, why? I have seen it, noticed 15 updated files, and decided to postpone the review until I have time for this.

[umputun]

the branch is a bit behind master (missing #245 and #247), would be good to rebase for a clean diff.

main issue I see — inconsistent forward-health-checks value parsing across providers:

• docker (docker.go:176-178): key presence alone enables it, so reproxy.forward-health-checks=false would still enable forwarding
• static (static.go:40): any non-empty string enables it, same problem with "false" or "no"
• consul-catalog (consulcatalog.go:195): properly checks for "true", "yes", "1"
• file (file.go:88): YAML boolean, works correctly

docker should follow the existing keep-host pattern which inspects the value. static should also handle "false"/"no" correctly.

minor: shouldForwardHealthChecks in health.go calls Match() on every /ping and /health request, then matchHandler calls it again later for the same request. not a big deal given health check frequency, just noting it.