Added some primitives, fixed some bugs (#111)

* added vector compress

* fixed missing return statement in compress

* added masked compare_as_imask

* changed avx512vbmi2 to avx512_vbmi2 in lscpu_flags

* changed signature of mask_equal_as_imask to get an imask instand of a mask

* added vector compress

* fixed missing return statement in compress

* added masked compare_as_imask

* changed avx512vbmi2 to avx512_vbmi2 in lscpu_flags

* changed signature of mask_equal_as_imask to get an imask instand of a mask

* Added popcnt

* Fixed typo for popcount (neon)

* Fixed std::popcount usage

* added capability to have multiple overloads with proper ambiguity check

* lint + flags fix for to_mask (sse)

* fixed redefinition check

* fixed potential segfault through to_array

* fixed neon imask_storeu

* added popcnt for smaller datatypes

* fixed typo

* added crazy-max ghaction-setup-docker to build-and-test.yml to enable multi-platform builds

* added install-qemu to arm build

* Added binary_mask_and for avx512

* fixed kand for avx512

* Added specific named output in tsl::type_name for cstdint types

* Added mask_sub

* Added support for blend on ARM Neon

* Reimplemented to_integral for NEON

* Update tsl_config.py

Changed pathlib.os.sep to os.sep

* Update misc.yaml

Changed implementation for blend on neon (now using bit select instead of nand / or)

* removed trailing space

* fixed some minor issues

* Restructured and improved CI/CD pipeline

---------

Co-authored-by: Alexander Krause <[email protected]>