i386: whack LINT-NO* kernels

[mjg]

Summary

there is nothing i386-specific about them and equivalent configs already build for amd64.

[emaste]

I suspect these are not telling us anything beyond what we get from amd64/arm64.

[imp]

These will tell us when we have i386-specific drivers that break the NOINET build. This used to be a frequent occurrence (monthly or so), but it's been maybe 10 years since it was that common.
There might also be 32-vs-64 bit issues, but the set of such issues that are unique to this environment is small (we may have had one or two detected in the 15-odd years we've had these).
I agree with ed: that set is (a) small, (b) fairly static and (c) likely not worth having these in universe for. There are other kernels that guard against more breakage than this.
We should consider an infrequent CI job, though, that does make universe MAKE_ALL_KERNELS=t though, but that's beyond the scope of this review... Even for that, these would provide little value.

[emaste]

i386-specific drivers that break the NOINET build

Ah, that is a good point and one I hadn't really thought about. I agree though that there's unlikely to be much change over time here.

[mjg]

I have never seen a breakage caught by any of these kernels which did not affect amd64.

[imp]

In D38700#880916, @mjg wrote:

I have never seen a breakage caught by any of these kernels which did not affect amd64.

Agreed. They used to provide value before we eliminated a number of old 10M drivers and others that were i386 specific. Now that they are greatly reduced, it's been maybe 10 years since I've had these catch anything. After they were first added, I'd see issues with my changes on a monthly or so basis.