[fhgffy]

CI builds with -Werror=switch-enum fail after #5762 because has_no_children() explicitly handles only array and object. A default branch does not suppress this warning.

Add the other eight value_t labels to the existing return true branch and regenerate the single header. The behavior stays the same. This fixes the warning reported on #5760 and complements the separate ObjectType fix in #5767.

Validation with MinGW GCC 15.1:

• The existing tests/abi/diag/diag_off.cpp fails before the patch with -std=c++11 -Wswitch-enum -Werror=switch-enum, using either header mode. It passes after the patch in C++11 and C++20.
• CMake builds with the same warning flags pass. Split headers: ABI diagnostics, allocator, regression2 and the test-data fixture, 4/4. Single header: ABI diagnostics, 1/1.
• Official amalgamation, pinned AStyle formatting and git diff --check pass.

The full suite was not rerun; the separate rbegin() build failure remains until #5767 lands.

• The changes are described in detail, both the what and why.
• If applicable, an existing issue is referenced (discussion on Fix NUL bytes in UBJSON/BJData high-precision numbers #5760).
• If applicable, a fixed OSS-Fuzz issue is referenced (not applicable).
• Code coverage remained at 100% (not measured locally; existing compile check used, no new executable path).
• If applicable, documentation is updated (no API or behavior change).
• The source code is amalgamated using the official generator.

[nlohmann]

Looks good to me.

[nlohmann]

Thanks!