urlEncode doesn't properly handle < 0x10

[unknownbrackets]

When trying to urlEncode a perfectly reasonable string containing newlines (%0a), they are encoded as %a which is incorrect and isn't properly parsed by other clients.

This is also affecting tabs. Other non-printables are probably unlikely to be used in URL encoding.

In my case I was trying to encode a multi-line string to place in a fragment for a link to translate.google.com. When properly encoded it happily accepts these strings.

I suppose the problem is here:
https://github.com/rejectedsoftware/vibe.d/blob/c8a03f93639ada7a307ea7dd3db0db909275a646/source/vibe/textfilter/urlencode.d#L44

-[Unknown]

[s-ludwig]

Fixed. The %a was not even parsed correctly by urlDecode() (and how should it, of course).

There are also some unit tests now. Reminded me that one big task left for 0.8 is to get a proper test suite and an auto tester up...