Documentation fixes

[yixuan]

This pull request fixes some mistakes in the previous documentation, and reflects recent API changes.

1. AVX2 now also has argsort, argselect, and key-value sort.
2. Adds the missing argument k in qselect and partial_qsort.
3. The version of argsort that returns a std::vector has been removed. Similar for argselect.
4. Fixes the template parameters in avx512_qsort_kv.
5. The sample code needs to include some xss-common-* headers to compile.

[r-devulap]

LGTM. Thanks for fixing!