Update langchain instrumentation (update to latest semantic conventions, bug fixes, update semantic conventions version, etc.)

[rads-1996]

Description

Update LangChain instrumentation to latest GenAI semantic conventions

• Added provider value mapping to match semantic conventions
• Normalized roles (human→user, ai→assistant, function→tool)
• Dropped [ChatOpenAI]/[ChatBedrock] allowlist; added ["model"] lookup so Gemini, Vertex, Anthropic, etc. emit spans
• Fixed [max_tokens] fallback for non-OpenAI providers
• Dropped fabricated "unknown"/"stop" finish reasons;
• Lowered tool calls/responses and reasoning blocks into spec message parts
• Bumped util-genai semconv pin to ~= 0.63b1; replaced magic strings with codegenned constants
• Deleted unused span_manager.py

Follow-up PR

• Missing inference attributes already exposed by LangChain: top_k, choice.count, stream, output.type, tool.definitions, system_instructions, server.address/port, cache & reasoning token counts, conversation.id

Fixes part of #126

Type of change

Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

Checklist

See CONTRIBUTING.md
for the style guide, changelog guidance, and more.

• Followed the style guidelines of this project
• Changelog updated if the change requires an entry
• Unit tests added
• Documentation updated

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Updates GenAI semantic convention usage and improves LangChain GenAI instrumentation to emit spec-conformant provider/role/message attributes (including Gemini and tool/function message handling).

Changes:

• Replaced hard-coded GenAI attribute keys with opentelemetry-semantic-conventions constants and bumped the semconv dependency.
• Added robust message conversion utilities (role normalization, tool calls/results, reasoning parts) and provider normalization.
• Updated LangChain callback handler and tests to reflect spec role/provider names and new span emission behavior (Gemini, finish reasons).

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
util/opentelemetry-util-genai/src/opentelemetry/util/genai/_inference_invocation.py	Switch reasoning token attribute to semconv constant.
util/opentelemetry-util-genai/src/opentelemetry/util/genai/_agent_invocation.py	Switch agent/version + cache token attributes to semconv constants.
util/opentelemetry-util-genai/pyproject.toml	Bump semantic conventions dependency to access GenAI constants.
instrumentation/opentelemetry-instrumentation-genai-langchain/src/opentelemetry/instrumentation/genai/langchain/utils.py	Introduce provider normalization + structured message conversion for roles/tool calls/reasoning.
instrumentation/opentelemetry-instrumentation-genai-langchain/src/opentelemetry/instrumentation/genai/langchain/callback_handler.py	Use new conversion utilities; expand model/provider support; improve finish_reason extraction.
instrumentation/opentelemetry-instrumentation-genai-langchain/tests/test_llm_call.py	Update assertions for spec roles/providers; add Gemini + FunctionMessage mapping test.
instrumentation/opentelemetry-instrumentation-genai-langchain/tests/test_callback_handler.py	Add tests for finish_reason defaults and propagation in output conversion.
instrumentation/opentelemetry-instrumentation-genai-langchain/src/opentelemetry/instrumentation/genai/langchain/span_manager.py	Remove legacy span manager implementation.

[opentelemetry-pr-dashboard]

This PR has review comments. Review suggestions, whether from maintainers or automated reviewers, aren't always correct or required. Please evaluate each comment on its merits, then make sure each thread has a clear outcome.

For example, link to the commit if you applied a suggestion, explain why it wasn't applied, or ask a follow-up question.

Automation flags a PR for human review once every review thread has a reply or is marked as resolved.

Status across open PRs is visible on the pull request dashboard.