Bridges based on bridgev2 no longer expose a metrics endpoint, so the
playbook was publishing a Traefik route which could only ever yield an
HTTP 502.

Related to https://github.com/spantaleev/matrix-docker-ansible-deploy/issues/5487

Part of adopting a uniform naming policy for bridge variables,
where the variable prefix matches the role directory name.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Auto-generate the provisioning shared secret (to enable the provisioning
API), route the whole bridge HTTP port via Traefik under
`<matrix-fqn>/bridges/meta-messenger`, and populate
appservice.public_address, reusing the matrix_bridges_exposure_*
mechanism. The labels template gate is widened so the exposure router is
emitted even when metrics are disabled (the exposure router reuses the
existing appservice Traefik service on port 29319).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Without this, we get a warning message from Traefik:

> INF Could not create a router for the container: too many services providerName=docker

Possibly due to the multiple services defined there without a single
explicitly-defined router.

Signed-off-by: Suguru Hirahara <acioustick@noreply.codeberg.org>

Related to: https://github.com/mautrix/facebook/issues/332

Fixes: https://github.com/spantaleev/matrix-docker-ansible-deploy/issues/3138