runtime: audit all assembly functions

[minux]

It seems we translated asm_ppc64x.s to asm_arm64.s, however,
at least we failed to replace ppc64x's closure environment register
(R11) with ours (R26).

[minux]

I did a first pass today, fixed some bugs, but I'm still
not sure all the assembly functions are correct.

but this issue is not as important as it was.

[minux]

Two more stack corruption bugs found in runtime/sys_linux_arm64.s.
(minux@4ea45ea)
After that, strings.test can pass a few tests:

$ ./strings.test -test.v
=== RUN TestCompare
--- PASS: TestCompare (0.00s)
=== RUN TestCompareIdenticalString
--- PASS: TestCompareIdenticalString (0.00s)
=== RUN TestCompareStrings
--- PASS: TestCompareStrings (0.04s)
=== RUN TestReader
--- PASS: TestReader (0.00s)
=== RUN TestReadAfterBigSeek
--- PASS: TestReadAfterBigSeek (0.00s)
=== RUN TestReaderAt
--- PASS: TestReaderAt (0.00s)
=== RUN TestReaderAtConcurrent
--- PASS: TestReaderAtConcurrent (0.00s)
=== RUN TestEmptyReaderConcurrent
--- PASS: TestEmptyReaderConcurrent (0.00s)
=== RUN TestWriteTo
--- PASS: TestWriteTo (0.00s)
=== RUN TestReplacer
--- PASS: TestReplacer (0.00s)
=== RUN TestPickAlgorithm
--- PASS: TestPickAlgorithm (0.00s)
=== RUN TestWriteStringError
--- PASS: TestWriteStringError (0.00s)
=== RUN TestGenericTrieBuilding
--- PASS: TestGenericTrieBuilding (0.00s)
=== RUN TestFinderNext
--- PASS: TestFinderNext (0.00s)
=== RUN TestFinderCreation
SIGILL: illegal instruction
PC=0x42087aad80

goroutine 1 [running]:

x0      0x0
x1      0x1
x2      0x2
x3      0x0
x4      0x4208022000
x5      0x4208000150
x6      0x48e8c
x7      0x4208000150
x8      0x0
x9      0x1
x10     0x3b9aca00
x11     0x7
x12     0x4be60
x13     0x0
x14     0x0
x15     0x0
x16     0x4208000750
x17     0x0
x18     0x0
x19     0x0
x20     0x0
x21     0x0
x22     0x0
x23     0x0
x24     0x0
x25     0x0
x26     0x0
x27     0x356333
x28     0x4208000120
x29     0x0
lr      0x42087aad80
sp      0x4208022a80
pc      0x42087aad80
fault   0x0