Added benchmark script, enabled ARMASM for Cortex-M3,4,7,33

[danielinux]

Adds a benchmark.sh script to compare ciphers & options over footprint, runtime memory, boot time.

Example output (in .md format): STM32H7 at 320MHz with ~100KB application image:

Name	Configuration	Bootloader size	Stack size	Image header size	Boot time
SHA2 only	SIGN=NONE	8776	1216	256	.388
SHA384 only	SIGN=NONE HASH=SHA384	18048	3760	256	.690
SHA3 only	SIGN=NONE HASH=SHA3	8472	1216	256	1.142
SHA2 only,small	SIGN=NONE NO_ASM=1	5564	1216	256	.475
rsa2048	SIGN=RSA2048	15176	17568	512	.518
rsa3072	SIGN=RSA3072	15316	12288	512	.670
rsa4096	SIGN=RSA4096	15460	18064	1024	.866
rsa4096 with sha384	SIGN=RSA4096 HASH=SHA384	24796	18064	1024	1.170
ecdsa256	SIGN=ECC256	21836	7632	256	.584
ecdsa384	SIGN=ECC384	21428	11216	512	2.535
ecdsa521	SIGN=ECC521	25104	8288	512	3.750
ecdsa256 with small stack	SIGN=ECC384 WOLFBOOT_SMALL_STACK=1	22184	5880	512	2.553
ecdsa256 with fast math	SIGN=ECC384 SP_MATH=0	21428	11216	512	2.535
ecdsa256, no asm	SIGN=ECC256 NO_ASM=1	14416	7632	256	3.347
ecdsa384, no asm	SIGN=ECC384 NO_ASM=1	15900	11216	512	11.516
ecdsa521, no asm	SIGN=ECC521 NO_ASM=1	14652	8288	512	25.735
ecdsa384 with sha384	SIGN=ECC384 HASH=SHA384	30704	11216	512	2.840
ed25519 with sha384, small	SIGN=ED25519 HASH=SHA384 NO_ASM=1	10976	5000	256	10.059
ed25519 fast	SIGN=ED25519 NO_ASM=0	27684	5000	256	9.242
ed448	SIGN=ED448	17432	4578	512	42.172
ML_DSA-44	SIGN=ML_DSA ML_DSA_LEVEL=2 IMAGE_SIGNATURE_SIZE=2420 IMAGE_HEADER_SIZE=8192	19520	19544	8192	.585
ML_DSA-65	SIGN=ML_DSA ML_DSA_LEVEL=3 IMAGE_SIGNATURE_SIZE=3309 IMAGE_HEADER_SIZE=8192	20160	19544	8192	.693
ML_DSA-87	SIGN=ML_DSA ML_DSA_LEVEL=5 IMAGE_SIGNATURE_SIZE=4627 IMAGE_HEADER_SIZE=12288	20800	19544	12288	.871
LMS 1-10-8	SIGN=LMS LMS_LEVELS=1 LMS_HEIGHT=10 LMS_WINTERNITZ=8 IMAGE_HEADER_SIZE=4096 IMAGE_SIGNATURE_SIZE=1456	11460	1320	4096	1.398
XMSS-SHA2_10_256'	XMSS_PARAMS=XMSS-SHA2_10_256 SIGN=XMSS IMAGE_SIGNATURE_SIZE=2500 IMAGE_HEADER_SIZE=8192	26952	9352	8192	1.205

[dgarske]

@danielinux wow! Hands-clapping!

[dgarske]

Tabs vs spaces? Looks like a few places

[dgarske]

"On a number of targets, algorithm may be optimized automatically to use assembly
optimizations." -> "Some targets support assembly optiomizations by default."?

[dgarske]

Are you planning to publish the benchmark.sh results in a CI job like the markdown you posted in the PR description?

[danielinux]

Yes that's the plan. Currently busy rebuilding my on-site infrastracture, I will eventually provide a dedicated jenkins node to run the benchmark.