Add server attributes to Vertex AI spans

[aabmass]

Description

Add server.address and server.port semconv attributes to Gen AI spans.

Type of change

Please delete options that are not relevant.

• New feature (non-breaking change which adds functionality)

How Has This Been Tested?

Please describe the tests that you ran to verify your changes. Provide instructions so we can reproduce. Please also list any relevant details for your test configuration

• Added new test case for parsing
• Updated vcr tests

Does This PR Require a Core Repo Change?

• Yes. - Link to PR:
• No.

Checklist:

See contributing.md for styleguide, changelog guidelines, and more.

• Followed the style guidelines of this project
• Changelogs have been updated
• Unit tests have been added
• Documentation has been updated

[lmolkova]

nit:
consider using 443 as a default (unless Vertex has grpc endpoint and uses a different port by default). If virtually all vertex spans would have port 443, let's just default to it and reduce telemetry volume a bit.

We have this clause for azure ai

server.port | int | GenAI server port. [5] | 80; 8080; 443 | Conditionally Required If server.address is set.

https://github.com/open-telemetry/semantic-conventions/blob/d9f3e97a6604ac0b451550c467b3f9cb1d1f492b/docs/gen-ai/azure-ai-inference.md?plain=1#L30

[aabmass]

If virtually all vertex spans would have port 443

Yes that's the case, and the suggestion makes sense to me. But isn't this special case for azure making server.port incompatible with the "parent type" GenAI span?

Conditionally Required If server.address is set.

[lmolkova]

we can add a note there that system-specific conventions can provide a default value. We can also add a small doc for vertex ai and mention default port there.

[lmolkova]

Left a minor suggestion to not populate port if it's 443, but LGTM