riscv: implement rvv optimized tan operator

[ihb2032]

This is a follow-up to #6436. I've applied the macro fix as suggested by @nihui to exclude xtheadvector, ensuring precision while keeping the optimization for standard RVV 1.0.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.90%. Comparing base (d1eb400) to head (e793dac).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #6460      +/-   ##
==========================================
+ Coverage   95.69%   95.90%   +0.20%     
==========================================
  Files         844      844              
  Lines      264779   265697     +918     
==========================================
+ Hits       253393   254821    +1428     
+ Misses      11386    10876     -510     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[nihui]

vfloat32m8_t operator()(const vfloat32m8_t& x, const size_t& vl) const
{
#if __riscv_xtheadvector
    // fallback
#else
    // rvv opt
#endif
}

[nihui]

Thanks for your contribution !