Match all MIPS and RISC-V MACHINE_ARCH values

[brooks]

Summary

The simple equality checks missed most possible values of MACHINE_ARCH
for MIPS and RISC-V. Compare to MACHINE_CPUARCH instead.

Use a glob for powerpc64 owing to powerpc64le.

[brooks]

There might be something to be said for a LINKER_FEATURES=lto in bsd.compiler.mk, but I'm not sure that we need to generalize from one example.

[brooks]

amd64 CI run https://cirrus-ci.com/task/5036030833197056

[brooks]

• give powerpc64 the match treatment as well

[brooks]

I remember that powerpc64le is on it's way so I've given powerp64 the same treatment.

[jrtc27]

usr.bin/gh-bc/Makefile

57–59

57–59

Though probably better as this. Also, is 32-bit powerpc definitely ok, or just not tested these days? Seems surprising that powerpc64 would be affected but not powerpc...

[brooks]

• Fix powerpc64 case.
• Use MACHINE_CPUARCH for mips and riscv

[brooks]

I'm likewise skeptical that any powerpc works

[brooks]

• Remove nonsensical !s

[jrtc27]

OOI do we know what the actual problem is? The comment is frustratingly light on detail so without trying it myself it's hard to have any idea where one would even look for a bug.

(and also extrapolate to other architectures)

[emaste]

needs rebase