Keep tool and agent names on GenAI semconv execute_tool and invoke_agent spans - #26394

# master at 59dac6f
$ python repro_26316.py    # the snippet from #26316
'get_weather' (TOOL) -> name='execute_tool' attrs=['gen_ai.operation.name', 'gen_ai.tool.call.arguments', 'gen_ai.tool.call.result']
'travel_planner' (AGENT) -> name='invoke_agent' attrs=['gen_ai.operation.name']
'my_pipeline' (CHAIN) -> name='CHAIN' attrs=['gen_ai.operation.name']
$ python repro_agent_run.py    # the same snippet with the one span pydantic_ai/autolog_v2.py:97 starts: Agent.run, typed AGENT
'Agent.run' (AGENT) -> name='invoke_agent' attrs=['gen_ai.operation.name']
$ python -m pytest tests/strands/test_semconv_probe.py -q -s | grep '^name='    # scratch test, not in this PR: a Strands agent with one tool, semconv export on
name='execute_tool' tool.name='sum' agent.name=None
name='invoke_agent' tool.name=None agent.name='agent'
$ python -m pytest tests/strands/test_semconv_probe_modelid.py -q -s | grep '^name=.*invoke'    # also scratch: the same agent with a model_id, as a default Agent() has
name='invoke_agent us.anthropic.claude-sonnet-4' tool.name=None agent.name='agent' model='us.anthropic.claude-sonnet-4'

# with this change
$ python repro_26316.py
'get_weather' (TOOL) -> name='execute_tool get_weather' attrs=['gen_ai.operation.name', 'gen_ai.tool.call.arguments', 'gen_ai.tool.call.result', 'gen_ai.tool.name']
'travel_planner' (AGENT) -> name='invoke_agent travel_planner' attrs=['gen_ai.agent.name', 'gen_ai.operation.name']
'my_pipeline' (CHAIN) -> name='CHAIN' attrs=['gen_ai.operation.name']
$ python repro_agent_run.py
'Agent.run' (AGENT) -> name='invoke_agent Agent.run' attrs=['gen_ai.agent.name', 'gen_ai.operation.name']
$ python -m pytest tests/strands/test_semconv_probe.py -q -s | grep '^name='
name='execute_tool sum' tool.name='sum' agent.name=None
name='invoke_agent agent' tool.name=None agent.name='agent'
$ python -m pytest tests/strands/test_semconv_probe_modelid.py -q -s | grep '^name=.*invoke'
name='invoke_agent agent' tool.name=None agent.name='agent' model='us.anthropic.claude-sonnet-4'

$ python -m pytest -q -n 4 -p no:cacheprovider tests/tracing/export/genai_semconv/test_translator.py tests/anthropic/test_anthropic_genai_semconv_converter.py \
    tests/bedrock/test_genai_semconv_converter.py tests/gemini/test_gemini_genai_semconv_converter.py \
    tests/openai/test_genai_semconv_converter.py tests/tracing/otel/test_span_translation.py
265 passed, 5 warnings in 23.56s
# baseline at 59dac6f: 257 passed, so the delta is the 8 new cases
# translator.py reverted, tests kept: test_translator.py gives 6 failed, 32 passed (test_full_tool_span,
# test_full_agent_span, both *_is_kept cases, the two updated test_span_name_and_kind rows)
$ pre-commit run --files mlflow/tracing/constant.py mlflow/tracing/export/genai_semconv/translator.py tests/tracing/export/genai_semconv/test_translator.py
ruff.....................................................................Passed
format...................................................................Passed
unresolved-import........................................................Passed
clint....................................................................Passed
typos....................................................................Passed
# every other hook Passed or Skipped; ruff 0.16.8 from the lint group, Python 3.12.3
# not run: the full test suite, and Semantic Kernel or pydantic-ai end to end (span names read from function_tracer.py:49 and autolog_v2.py:97)