PrmDb: use canonical serialized size for record size

[hidirektor]

Related Issue(s)	#4716
Has Unit Tests (y/n)	y
Documentation Included (y/n)	n
Generative AI was used in this contribution (y/n)	y

---

Change Description

Updates Svc/PrmDb to use the canonical F Prime serialized size format for the file record size field.

• Replaces raw U32 record-size serialization/deserialization with serializeSize(...) / deserializeSize(...)
• Aligns record-size read/write checks with sizeof(FwSizeStoreType)
• Adds a new FW_HAS_16_BIT == 1 regression test that:
  • writes a 16-bit parameter value,
  • saves the parameter file,
  • verifies the on-disk layout by deserializing the saved bytes,
  • reloads the file and confirms round-trip behavior

Rationale

This fixes the parameter database file format so it is not tied to host U32 assumptions and is more robust across configurations with different type sizes.

The added regression test protects against future format regressions, especially in non-default build configurations such as 16-bit payloads.

Testing/Review Recommendations

• Review the record-size serialization path in Svc/PrmDb/PrmDbImpl.cpp
• Review the new 16-bit regression coverage in Svc/PrmDb/test/ut/PrmDbTester.cpp
• Confirm the read/write error expectations now use FwSizeStoreType-based sizing
• Validate that the round-trip test still passes in FW_HAS_16_BIT == 1 builds

Future Work

• Add broader coverage for additional non-default size/type configurations if needed
• Consider adding a dedicated file-format compatibility test for other payload widths

AI Usage (see policy)

AI was used for review support.

[LeStarch]

Why did this change? I believe this loops through each record. Size was in case 2, so my expectation that case 3 remain PARAMETER_ID

[hidirektor]

That may seem like an odd change at first, but this test fails by read() call index (m_waits), not by field name. With the size-format update, case 3 now hits a record-size read at the next record boundary, so PrmReadError::RECORD_SIZE_SIZE is the correct expectation

[LeStarch]

To fix the formatting, you can run this command:

git diff --name-only devel...HEAD | fprime-util format --stdin