[RFC] removed webpath set to null on unset resolver

[c33s]

Q	A
Branch?	2.0 would maybe also make sense to backport it
Bug fix?	yes (for me yes)
New feature?	no
BC breaks?	yes (kind of, if someone buld upon this fallback)
Deprecations?	no
Tests pass?	?
Fixed tickets	#1132
License	MIT
Doc PR

alredy commented in https://github.com/liip/LiipImagineBundle/issues/1132#issuecomment-524867253 but i thought a PR is more visible to get comments.

haven't tried to run or fix tests because first it makes sense to know if you would merge this.

[michellesanver]

Thanks for this PR! I would definitely merge this bugfix and release in 3.0 as it is indeed a slight BC break. I'd be super happy if you go ahead and fix the related tests.

[c33s]

sounds great. will have a look at the tests this week

[dbu]

i don't think we can simply remove the default resolver. maybe we could add it in the extension class instead of the configuration class to avoid the problem when merging configurations?