fixed double decoding in urldecode

[akrainiouk]

Current falco implementation of urldecode essentially applies url decoding twice, once by applying url.PathUnescape and then by calling url.QueryUnescape. In addition %25 (escaped "%") gets yet another extra decoding.

Fastly documentation of urldecode is a bit misleading:

For example, urldecode({"hello%20world+!"}); and urldecode("hello%2520world+!"); will both return "hello world !".

This makes an impression that both %20 and %2520 should be decoded to space character
However in reality it means to demonstrate is that two different string literals "..." and {"..."} are interpreted differently. In the second example %25 is decoded to '%' by the paraser leading to the same original value passed to urldecode as in the first example.

The confusion is also likely caused by misinterpreting of quoted literals in falco which is apparently fixed in PR-256

[richardmarshall]

Thanks, this looks good.