Support real regex rules for rewrite and proxy middleware

[lammel]

Based on a real-world application using a workaround to ignore a part of the URL using .+? is not working anymore.
Using /prefix/*/*is also problematic due to the greedy nature of our regex conversion.

The behavior for handling regex (using QuoteMeta now) has been changed with PR #1630.

With this PR the rewrite middleware will be enhanced to also allow real regular expressions with capture groups.
The existing Rules can still be used and will be converted to the now public RegexRules. Both can be used together, as long as the resulting regex is not exactly the same as a regex generated from a rules section.

Example for usage:

e.Pre(RewriteWithConfig(RewriteConfig{
		Rules: map[string]string{
			"^/a/*":     "/v1/$1",
			"^/b/*/c/*": "/v2/$2/$1",
			"^/c/*/*":   "/v3/$2",
		},
		RegexRules: map[*regexp.Regexp]string{
			regexp.MustCompile("^/x/.+?/(.*)"):   "/v4/$1",
			regexp.MustCompile("^/y/(.+?)/(.*)"): "/v5/$2/$1",
		},
	}))

A separate PR for documentation will be created in https://github.com/labstack/echox

[codecov]

Codecov Report

Merging #1767 (51df801) into master (7c8592a) will increase coverage by 0.01%.
The diff coverage is 83.33%.

@@            Coverage Diff             @@
##           master    #1767      +/-   ##
==========================================
+ Coverage   89.72%   89.74%   +0.01%     
==========================================
  Files          32       32              
  Lines        2667     2672       +5     
==========================================
+ Hits         2393     2398       +5     
  Misses        175      175              
  Partials       99       99              

Impacted Files	Coverage Δ
middleware/rewrite.go	76.47% <66.66%> (+3.13%)	⬆️
middleware/middleware.go	100.00% <100.00%> (ø)
middleware/proxy.go	62.35% <100.00%> (+1.37%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 7c8592a...51df801. Read the comment docs.

[aldas]

Other than that comment copy paste looks good to me.

p.s. this is my observation
And we really should use more table driven tests and even maybe refactor some old tests to tables when touching different parts. Those sequential page length tests could have side-effect and pass only because some earlier scenario/testcase already changed some internal state that next test case touches by reusing objects.

[lammel]

@aldas Feel free to merge, if Goland is happy now too.
I'll update the release notes, once it's merged.