Make blib be same author-side as for install-side for better dev/testing

[mohawk2]

A major benefit of this is it allows dev of complex distros like PDL, which is in fact why I made these changes.

[ingydotnet]

Can you elaborate on what this does and the problem it solves.

I don't have time at the moment to dive into this.

[mohawk2]

Certainly, and once you're happy with it, I'll make a suitable doc patch too.

Use-case: PDL. I am making the various *.pd files instead be Inline::Pdlpp modules. There are a number of them, and some of the inline* tests (quite reasonably) set Inline's config dir to suitable places. A naive I::Pdlpp module will use that and rebuild itself, causing delays. Additionally, since the "dyna" type module uses a SO out of blib, it only seems correct that author-side testing will be working off the same environment as the hapless user.

Therefore, this PR adds a specific build_inline target, which hangs off not pure_all but dynamic, because Inline doesn't currently support static. For each installable Inline module, it does:

• makeblibproxy which creates blib/lib/PDL/*/Inline.pm as a proxy module (allowing Inline to rebuild if necessary)
• runs the rebuild
• makeblibdyna which creates blib/lib/PDL/*/Inline.pm as a dyna module

At the end, as currently, it runs fixblib, which now instead of using link, uses File::Copy::cp for greater portability.

I just checked all this works with a make disttest, and it does. An additional benefit is that there is now no need to keep the stub module yourself, since it's correctly placed in blib at the right time ready for installing, and obviously already gets put in the distdir - that could also safely be omitted.

[mohawk2]

(The above is only done on EUMM at the moment but I'm sure it wouldn't take much to port the logic to MB, which would then allow deleting the distdir_inline code)

[ingydotnet]

This seems like good stuff, but I still need time to think through and digest it. Will try to do this soon. Thanks for the work.