Rework non-empty-string assertions

[herndlm]

This does a couple of things

• Removes the hacky non-empty-string expression re-usage and replaces it with type specifying and calls to $this->typeSpecifier->create (I hope this closes False positive on Assert::upper() always evaluates as true #85)
• Removes the expression adaption for nullOr* assertions and replaces it with TypeCombinator::addNull and calls to $this->typeSpecifier->create. This should close Assert::nullOrStringNotEmpty() with string|null will always evaluate to true #33.
• Tries to detect the type from SpecifiedTypes a bit smarter by also looking at the sureNotTypes. This retains non-empty-string via nullOr* and all* better
• Adds integration and many other tests, e.g. some cases of all* and nullOr* almost everywhere, where it makes sense
• Fixes some of the non-empty-string resulting assertions that expect a string as argument (e.g. contains or notWhitespaceOnly)

I know this is a big one, I'm happy to split it up or whatever makes sense. Some of the non-empty-string things seem to be a bit intertwined though.

[VincentLanglet]

Thanks for the pretty quick fix.

[herndlm]

@VincentLanglet are you affected by a bug, e.g. #85? Maybe you can help me reproducing it or adding a regression test :)

[VincentLanglet]

@VincentLanglet are you affected by a bug, e.g. #85? Maybe you can help me reproducing it or adding a regression test :)

The following code

        Assert::stringNotEmpty($url);
        Assert::contains($url, '/'); // => Always true
        Assert::startsWith($url, 'https://github.com/');  // => Always true

report two errors (with the phpstan strict plugin).

[herndlm]

Ok, I think I was successful in adding the regressiontest by using strict-rules. But it shows that the problem still persists, so I'm kind of stuck.
This PR then fixes/improves some nullOr* all* assertions and does some refactoring I guess, but it does not fix the referenced issues unfortunately. They seem to be coming from phpstan's ImpossibleCheckTypeStaticMethodCallRule and I'm not sure yet if they make sense or not. Might be a incompatibility with non-empty-string? Maybe Ondrej can help out, sorry..

From PHPStan's point of view, based on types, it makes sense I guess to think that e.g. upper does not add additional type value to something that is already non-empty-string because e.g. there was a "length is 3" check before.