Remove special behavior for rake

[JasonLunn]

Fixes #9244

[Edouard-chin]

I also encountered this code a few weeks ago and couln't figure out why it was needed. FWIW it was added in 20cdf22b844 but I can't find the original rubyforge discussion that led to this change.

[hsbt]

@JasonLunn Can you add test for this? I couldn't confirm this change works correctly.

[Edouard-chin]

@matthewd Found the rubyforge discussion. This is the reason this patch was applied

https://web.archive.org/web/20140515164351/http://rubyforge.org/pipermail/rubygems-developers/2006-July/002047.html

I filed a bug report about this this morning, but just wanted to
mention it here briefly. RubyGems 0.9.0 introduced a bug for
installing compiled extensions; in addition to doing the "make" and
"make install" steps of the build process, it now follows up with a
"make clean", which deletes the object files (.o) and the extension
shared library (.so). And, you know, that sort-of defeats the
purpose. ;)

and the patch was proposed here https://web.archive.org/web/20140515164338/http://rubyforge.org/pipermail/rubygems-developers/2006-July/002056.html

[JasonLunn]

Regression test added.

[JasonLunn]

@hsbt - can you re-run tests now that the linter has been appeased?

[JasonLunn]

Extra whitespace fixed and re-based, ready for re-testing

[JasonLunn]

Thanks, @hsbt!