Support for SME1 based ssymm_direct kernel for cblas_ssymm level 3 API

[matcraje]

This PR introduces a specialized kernel that utilizes ARM SME1 (Scalable Matrix Extension) capabilities to optimize the cblas_ssymm function.

[matcraje]

Looks like cmake builds are failing. I will check and modify the changes.

[matcraje]

The checks using 'make' are successful.
The checks using 'cmake' fail due to undefined symbols (Eg. undefined symbol '_ssymm_direct_alpha_betaLL_A64FX' etc).

@martin-frbg Do you know if I missed something?

[martin-frbg]

kernel/CMakeLists.txt has a second block of definitions starting around line 1100 that specifically handles DYNAMIC_ARCH builds - you need to add the equivalent lines for your ssymm_direct_alpha_betaLL etc. with an added ${TSUFFIX} there

[martin-frbg]

WoA build still fails with the

2025-09-30T05:58:18.3418812Z lld-link: error: duplicate symbol: __arm_tpidr2_save
2025-09-30T05:58:18.3419188Z >>> defined at kernel\CMakeFiles\kernel.dir\CMakeFiles\ssymm_direct_alpha_betaLU.c.obj

that we've come to understand happens when arm_sme.h is included unconditionally - please fix

Also please move your additions in interface/symm.c after the error handling (the line where xerbla is called if "info" is not zero) - this will probably require another #if defined CBLAS but protects your kernel from getting called with invalid arguments

[matcraje]

Thanks @martin-frbg for the comments. Earlier errors are now resolved.