Task: Implement Hotfix: CI builds red across all OSs on the payload decode fold and the ascot member

Apply the five fixes the story names, each verified by the class of its change, and raise one PR that carries all of them.

• Every fix is in the PR branch, with the class checks for code and ci changes run locally and recorded in the PR body.
• The build and ctest are green under linux-clang-debug-make, and the clang-19 and gcc probes on the two failing translation units are clean.
• The PR states plainly that the Windows fixes cannot be verified locally.

1. Read the failing CI logs on the red head and map each workflow to its compile diagnostic. Five independent defects came out of that.
2. Rename the ascot member to ascot_row in bond_instrument_data and update its call sites, so gcc stops rejecting the collision between the member name and its type name.
3. Rewrite decode_with_legs as the try_one_with_legs / try_each_with_legs helper pair, the shape decode_flat already uses, so clang 19 and AppleClang 16 stop crashing in the fold expression.
4. Disable C4251 for MSVC at the top of projects/CMakeLists.txt. Add the Windows-scoped -Wno-unused-function and /wd4505 for the generated domain.cpp in ores.ore/core/src/CMakeLists.txt.
5. Move -performance-avoid-endl out of the invalid Warnings: key and into Checks: in .clang-tidy. Add the dev-package and Qt install steps the Nightly Include Hygiene job was missing.
6. Prove each fix locally where the compiler is available: gcc 16.2.0 and clang 19.1.7 probes on the two failing translation units, and a clang-tidy run with the fixed config.
7. Run the class checks: full build, full ctest with a CDash experimental submission, the roundtrip check, both drift checks, the site build and the doc lint. The ctest run surfaced a sixth defect the build failures had masked: the bond future exporter test document still carries the pre-v17 shape. Fix the document and re-run.
8. Raise one PR carrying all six fixes, and state which of them Linux cannot verify.

The red head is e0a339a7. Every workflow fails, and every failure is a compile or configure error:

Three things limit what the local runs can prove, and the PR says so.

MSVC dies inside ores.platform, the first library of the build, so it never reaches the generated file. The /wd4505 flag and the Windows-clang -Wno-unused-function are therefore predictions from the diagnostic text, not observations.

The Nightly Include Hygiene job fails at configure, so it never reaches clang-tidy. The .clang-tidy defect is proven locally only, by running clang-tidy with the pre-fix and post-fix config.

The sixth defect is pre-existing on main. The test file and the XSD are byte-identical on both sides of the branch point, and the failing parse sits in xsdcpp-generated code the branch never touches. CI runs ctest after the build, so the defect would have turned every job red again the moment the build fixes landed.

The ORE domain bindings under ores.ore are xsdcpp output and the drift check regenerates them clean, so no generated file is edited by hand.

Manual QA scenarios (scaffolded via compass add test_scenario, run through the QA Validation Runner panel) that verify this task. Link new ones here as they're created; the scenario doc itself links back via its "Verifies task" field.

Six fixes sit on feature/implement-ci-builds-red-payload-fold-hotfix, one commit each.

• gcc: the ascot member is renamed ascot_row, with its call sites updated. A gcc 16.2.0 probe on the pre-fix header reproduces the CI diagnostic, and the fixed header compiles clean.
• clang: decode_with_legs is now the try_one_with_legs / try_each_with_legs helper pair. clang 19.1.7 on the pre-fix file reports PLEASE submit a bug report and exits 1; the fixed file exits 0.
• MSVC: C4251 is disabled for MSVC alone.
• Windows clang: the generated domain.cpp gets -Wno-unused-function, and MSVC gets /wd4505, both scoped to that one file.
• clang-tidy: the invalid Warnings: key is gone and the config parses.
• Nightly Include Hygiene: the job installs the dev packages and Qt6 before it configures.
• The bond future test document carries the three elements the v17 schema keeps on the trade side.

Class checks, all on the final tree: the full build under linux-clang-debug-make is clean; the full ctest passes 71 of 71 suites and the CDash experimental submission succeeds; ore_domain_roundtrip_check.py exits 0; the cmake-sources and codegen drift checks report no drift; the site build and the doc lint are clean.

The branch then rebased onto main 34ed6cf1, which carries the clang-format merge, and every class check was re-run on the rebased tree: the build is clean, ctest passes 71 of 71 with the CDash submission successful, both drift checks report no drift, the roundtrip check exits 0, the doc lint and the site build are clean, and the seven changed C++ files are clang-format clean under clang-format 21.1.8.

The first post-rebase ctest run was red, with 8 of 71 suites failing. All 117 failures carried one message, NATS connect failed: No server available for connection. The eventing integration suites need the environment's nats-server-brave_hopper unit, which was stopped at the time. With the unit up the run is 71 of 71. No failure was a code defect.

The first two acceptance criteria are met. The third is met in the PR body: the MSVC and Windows-clang fixes and the nightly Qt install cannot be verified on this machine, and a further Windows error may sit behind the first one.