Faster scalar inversions for P-384 and P-521 #5490

Job	Run time
Windows (shared, x86_64, windows-2022)	5m 37s
Clang Tidy	44s
Windows (static, x86_64, windows-2022)	4m 22s
Windows (amalgamation, x86_64, windows-2022)	4m 24s
Linux (gcc, shared)	1m 57s
Windows (shared, x86, windows-2022)	7m 50s
macOS (shared, xcode, macos-13)	9m 24s
Linux (gcc, amalgamation)	3m 32s
macOS (amalgamation, xcode, macos-13)	3m 58s
Linux (gcc, static)	1m 38s
macOS (shared, xcode, macos-14)	2m 22s
Linux (clang, shared)	2m 22s
Analysis (coverage, gcc, ubuntu-24.04)	26m 40s
Special (examples, gcc, ubuntu-24.04)	47s
macOS (amalgamation, xcode, macos-14)	2m 41s
Analysis (sanitizer, msvc, windows-2022, ninja)	18m 41s
Special (minimized, gcc, ubuntu-24.04)	15s
Analysis (sanitizer, clang, ubuntu-24.04)	11m 7s
Special (bsi, gcc, ubuntu-24.04)	47s
Cross (cross-i386, gcc, ubuntu-22.04)	2m 38s
Analysis (sanitizer, gcc, ubuntu-24.04)	17m 48s
Special (docs, gcc, ubuntu-24.04)	43s
Cross (cross-arm32, gcc, ubuntu-24.04)	9m 41s
Analysis (valgrind, clang, ubuntu-24.04)	25m 48s
Cross (cross-arm64, gcc, ubuntu-24.04)	6m 15s
Analysis (fuzzers, gcc, ubuntu-24.04)	6m 15s
Analysis (lint, gcc, ubuntu-24.04)	51s
Cross (cross-ppc64, gcc, ubuntu-24.04)	6m 6s
Analysis (format, gcc, ubuntu-24.04)	23s
Cross (cross-mips64, gcc, ubuntu-24.04)	6m 29s
Analysis (limbo, gcc, ubuntu-24.04)	56s
Cross (cross-android-arm64, clang, ubuntu-24.04)	49s
Cross (static, gcc, windows-2022, make)	5m 8s
Cross (cross-ios-arm64, xcode, macos-13)	7m 41s
	3h 26m 39s

Provide feedback