Add process autodiscovery infrastructure and support to krakend (#24238)

* process autodiscovery for krakend

Adds process-based autodiscovery for krakend, on top of the existing
container-based autodiscovery, using a CEL selector to match the
krakend process by name. Both flavors run as parametrized variants of
the same e2e test in a single environment: the process variant scopes
DD_AUTOCONFIG_EXCLUDE_FEATURES=docker to just that discovery invocation
(via the new ddev --env support) so that invocation's autodiscovery run
never learns about containers, letting the otherwise container-bound
krakend process be matched by process-based autodiscovery instead.

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

* assert

* comment

* time

* drop

* Gate get_e2e_discovery_metadata's process-AD env/caps behind a process flag

Only krakend needs the process-autodiscovery env vars and capabilities;
returning them unconditionally to all 8 callers grants unneeded caps to
7 unrelated integrations and, for ray, silently clobbers its own env_vars
via a shallow dict merge.

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

* Remove leading underscore from assert_discovery_used_expected_mechanism

AGENTS.md only permits a leading underscore on class instance methods
or non-obvious-behavior helpers in reusable library modules; this is a
plain module-level free function, so it shouldn't have one.

Environment: Datadog workspace

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

---------

Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> 5f35075

Author: github-merge-queue[bot]