[CUDA] Use qmv kernel for fp quantizations (ml-explore#3239)

Bump the minor version (ml-explore#3183)

Patch for multi device CUDA (ml-explore#3100)

Fix for NAX overflow. (ml-explore#3092)

remove thrust (ml-explore#3067)

Allow some non 2D inputs in qqmm (ml-explore#2981)

Fix warnings for the NAX build (ml-explore#2921)

Add Neural Accelerator Support (ml-explore#2772)

fix release builds (ml-explore#2746)

suppress gcc 10.1 warnings (ml-explore#2679)

* suppress gcc 10.1 warnings

* suppress gcc 10.1 warnings