balar: repair trace test setup and add basic contracts

[nab880]

Existing BalarTestCPU trace tests have broken skip decorators, stale fixture paths, and no cudaFree replay. Repair that setup and add basic allocation/free and wide-copy contracts using the existing driver. Isolate optional RISC-V/LLVM toolchain setup, shared statistic checks, and the direct-memory test topology.

This layer adds no DoorbellTestCPU and does not change MMIO runtime behavior.

Validation:

• Installed SST test runner discovery succeeds: 6 basic/smoke tests plus the 4 wire-header tests, with no duplicates.
• All 15 Python files parse; every Makefile source/distribution reference exists.
• No Doorbell driver imports, staging references, or build sources remain in this layer.
• git diff --check passed. Full CUDA/GPGPU-Sim integration was not run locally.

Incremental review PR in the contributor fork, based on balar-1-wire-header. Upstream branch protection prevents feature-branch bases there. After the preceding layer lands upstream, submit this branch against sstsimulator/sst-elements:devel; retain this fork PR as its incremental review record.

Stack (review order):

1. balar-1-wire-header
2. balar-2-test-support
3. balar-3-doorbell-driver
4. balar-4-compact-mmio