spdmd: add the support for platform composite attestation

[helloxiling]

• Add the composite attestation core for building detached Claims-Sets, submodule digests, and tag-602 Detached EAT bundles from collected SPDM evidence.
• Wire the optional D-Bus composite bundle interface behind the composite-attestation Meson feature.
• Define the vendor-neutral Lead Attester interface, and limit the upstream concrete backend to the software mock used for development and test.

Validation

• Standalone composite/mock attestation tests pass: 11 suites / 64 tests.
• QEMU emulation smoke test completed with the local test patch:
  • spdmd and bmcweb active
  • GET /redfish/v1/ComponentIntegrity/ exposed one synthetic responder
  • POST /redfish/v1/ComponentIntegrity/Actions/Oem/Oem.GetCompositeEATBundle returned 202
  • GET /redfish/v1/ComponentIntegrity/CompositeEATBundle/ returned Status: Ready
  • Generated bundle verified successfully: COSE_Sign1 ES384 signature, nonce/UEID/profile/submods/measurements, and submodule digest recomputation.
• E2E hardware validation with Microsoft OpenBMC and a real RoT acting as the lead attester.

[ratagupt]

Xiling , I really like the code. It is written very neatly and is well documented. I have completed my first pass and only have a few comments so far. I will do another round of review as well.

[helloxiling]

Xiling , I really like the code. It is written very neatly and is well documented. I have completed my first pass and only have a few comments so far. I will do another round of review as well.

@ratagupt Ratan, appreciate for your detailed review! Those comments are very helpful for the code enhancement. The latest patch addressed them. Please let me know if there are any questions or concerns.

[LeeTroy]

We had compiling issue when integrating with latest openbmc (boost 1.91).
It seems the ec doesn't process after cancel(ec), is it reasonable to use cancel() instead?