feat(scheduling): add endpoint attribute weight scorer

[nicole-lihui]

What type of PR is this?

/kind feature

What this PR does / why we need it:

Add label-producer and endpoint-attribute-weight-scorer.

The producer exposes Pod labels as endpoint attributes, and the scorer maps discrete attribute values to normalized routing scores. This supports static preferences for heterogeneous GPUs, zones, racks, and similar endpoint groups while keeping missing or unknown values eligible.

Which issue(s) this PR fixes:

Fixes #2201

Test plan:

• Unit tests cover configuration validation, score normalization, fallback behavior, and scorer category.
• Hermetic integration tests verify weighted GPU routing and eligibility of missing or unknown attributes.

Release note (write NONE if no user-facing change):

Add endpoint attribute weight scoring for configurable static routing preferences.

[elevran]

thanks @nicole-lihui !
Q: given that labels are already part of the Endpoint metadata state, what is the added value of making labels into Endpoint attributes? It can probably result in more focused/granular plugins, but I wonder what other reasons might be.