Enable SuperPMI replay for APX ISA testing

[BruceForstall]

Create a new runtime-coreclr superpmi-replay-apx pipeline that
does SuperPMI replay using AltJit with various APX enabling configuration
options set. This enables some amount of APX testing even without
hardware available to run the generated code.

Notes:

1. Create a new jit-replay-pipeline.yml template to share the
   superpmi-replay pipeline implementation with the existing pipeline.
2. The existing pipeline is designated the "standard" replay type and
   the new pipeline the "apx" replay type.
3. The APX pipeline does replays for Windows x64 and Linux x64.
4. The configurations tested are:

a. "EnableAPX=1"
b. "EnableAPX=1", "EnableApxNDD=1"
c. "EnableAPX=1", "JitStressRex2Encoding=1"
d. "EnableAPX=1", "JitStressPromotedEvexEncoding=1"
e. "EnableAPX=1", "JitStressRegs=4000"
f. "EnableAPX=1", "EnableApxNDD=1", "JitStressRex2Encoding=1", "JitStressPromotedEvexEncoding=1", "JitStressRegs=4000"

All configurations pass --altjit to superpmi.py to set the AltJit variables, as well as
setting "RunAltJitCode=0" which causes all enabled ISAs to be enabled.
5. Change superpmi to no fail altjit compiles when RunAltJitCode=0 is set. With RunAltJitCode=0, the JIT returns CORJIT_SKIPPED. We don't want superpmi to fail these compiles.
6. nit: fixed typo in superpmi of getStingConfigValue => getStringConfigValue

[BruceForstall]

@anthonycanino @DeepakRajendrakumaran @Ruihan-Yin @kunalspathak PTAL
cc @dotnet/jit-contrib

[Copilot]

Pull Request Overview

This PR introduces a new SuperPMI replay pipeline for APX ISA testing and updates existing pipelines and scripts to support a new "apx" replay type. Key changes include:

• Adding a new pipeline file (superpmi-replay-apx.yml) configured for APX replays.
• Updating the shared pipeline template and job definitions to accept a replayType parameter.
• Modifying the SuperPMI replay setup and replay scripts to handle "standard" and "apx" configurations.

Reviewed Changes

Copilot reviewed 8 out of 14 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
eng/pipelines/coreclr/templates/jit-replay-pipeline.yml	Adds new parameters (including replayType) for pipeline configuration.
eng/pipelines/coreclr/superpmi-replay-apx.yml	Introduces a new pipeline definition for APX replays.
src/coreclr/scripts/superpmi_replay_setup.py	Updates argument parsing to require and verify the replay type.
src/coreclr/scripts/superpmi_replay.py	Adjusts configuration logic to select between standard and APX settings.
eng/pipelines/coreclr/templates/superpmi-replay-job.yml	Updates job naming and artifact retrieval to incorporate replayType.
eng/pipelines/common/templates/runtimes/send-to-helix-step.yml	Adds a new parameter for passing the replay type to Helix.
eng/pipelines/coreclr/superpmi-replay.yml	Updates the pipeline to extend the new jit-replay-pipeline template with replayType.
eng/pipelines/coreclr/templates/run-superpmi-replay-job.yml	Revises the setup call and variable propagation to include replayType.

Files not reviewed (6)

• src/coreclr/scripts/superpmi-replay.proj: Language not supported
• src/coreclr/tools/superpmi/superpmi-shared/methodcontext.cpp: Language not supported
• src/coreclr/tools/superpmi/superpmi-shared/methodcontext.h: Language not supported
• src/coreclr/tools/superpmi/superpmi/icorjitinfo.cpp: Language not supported
• src/coreclr/tools/superpmi/superpmi/jitinstance.cpp: Language not supported
• src/coreclr/tools/superpmi/superpmi/jitinstance.h: Language not supported

Comments suppressed due to low confidence (1)

src/coreclr/scripts/superpmi_replay.py:22

• [nitpick] Using the argument name "-type" shadows the Python built-in function 'type'. Consider renaming it to "-replayType" (and updating references accordingly) to improve clarity.

parser.add_argument("-type", required=True, help="Type of replay (standard, apx)")

[anthonycanino]

I think we can fold c and d into a single configuration, and eventually may want to do away with them as separate stress modes entirely. The stress modes as separate variables was mostly due to implementation purposes.

[BruceForstall]

I think we can fold c and d into a single configuration, and eventually may want to do away with them as separate stress modes entirely. The stress modes as separate variables was mostly due to implementation purposes.

It wasn't clear to me how useful the independent stress variables are, or if it's valuable/necessary to test them independently. As the stress variables change/merge, we can adjust the testing. E.g., I presume we'd add JitEnableApxConditionalChaining=1 after #111072 is merged.

[BruceForstall]

Contributes to #110672

[kunalspathak]

Did you get a chance to do a dry run with this pipeline?
LGTM otherwise.

[BruceForstall]

Did you get a chance to do a dry run with this pipeline? LGTM otherwise.

I tested both the "standard" (existing) replay pipeline and the "apx" replay pipeline (by re-using the existing AzDO pipeline) before submitting this PR. After this is merged, I'll get a separate "apx" pipeline created and configured.

[BruceForstall]

/ba-g Unrelated test failures