SHA256 implementations based on Intel SHA Extensions

Summary:

• [Refactor] CPU feature detection logic for SHA256
• Add SHA256 implementation using using Intel SHA intrinsics
• Use immintrin.h everywhere for intrinsics
• Use __cpuid_count for gnu C to avoid gitian build fail.

This is a backport of Core PR13386 and PR13611

Depends on D1850

Test Plan:

make check

Reviewers: #bitcoin_abc, schancel

Reviewed By: #bitcoin_abc, schancel

Subscribers: teamcity

Differential Revision: https://reviews.bitcoinabc.org/D1851