fix: restore config management support for mapproxinator

[roicohen326]

Problem

mapproxinator:v1.3.0 adopted @map-colonies/config, which fetches configuration over HTTP at startup. The chart sets no CONFIG_* variables, so the library falls back to its default server URL and the container exits immediately:

ConfigError [httpGeneralError]: An error occurred while making the request
  at async initConfig (/usr/src/app/common/config.js:14:22)
  code: 'ECONNREFUSED', address: '127.0.0.1', port: 8080

This is currently crashlooping in qa and integration after the 2.3.1 version alignment bumped mapproxinator to v1.3.0. Neither environment is down — the previous v1.2.5 pods stay up because the new ones never pass readiness.

Cause

This support existed once. #85 added it on 25 Jun. #86 — the v2.0.0 MapProxy 6 rewrite — removed it one day later:

git log --oneline -- helm/templates/mapproxinator/mapproxinator-configmap.yaml
e9376a0 feat!: rewrite as MapProxy 6 image, v2.0.0 (supersedes mc-mapproxy) (#86)   <- removed
07f4b21 chore: helm config management (#85)                                        <- added

So 2.0.0, 2.1.0 and 2.1.1 all shipped without it. Chart 6.0.1 in mc-mapproxy is missing it too.

Change

Restores the #85 implementation in both configmaps — the init container and mapproxinator use separate ones, and both run the same image, so both need it.

One deliberate deviation from #85: version defaults to 'latest', not ''. Verified against the real image — an empty value fails schema validation:

'version' property must be equal to the allowed value
path: '{base}.version', context: { errorType: 'const', allowedValue: 'latest' }

serverUrl now matches the exporter-trigger convention (http://localhost:8080/api) rather than ''.

Defaults are unchanged in effect (offlineMode: false); consumers opt in.

Verification

Rendered env applied to mapproxinator:v1.3.0 running in-cluster:

{"level":"info","msg":"starting initMode"}
{"level":"info","msg":"initializing configuration"}     <- passes, previously threw ECONNREFUSED
...
ENOENT: no such file or directory, stat '/path/to/source/mapproxy.yaml'

Config init succeeds; the process proceeds to its normal FS provider step and stops only because the throwaway test pod had no mapproxy.yaml mounted. The chart mounts it in a real deployment.

Also verified:

• helm lint — 0 failed
• helm template renders all four keys in both configmaps
• both override paths work — per-service and global.configManagement.offlineMode=true (one value covering both containers)

Consumer follow-up

helm-charts then sets global.configManagement.offlineMode: true under mapproxy in serving-values.yaml and bumps the chart, across master / qa / integration.