Make converters injectable and respect culture

[JimBobSquarePants]

Prerequisites

• I have written a descriptive pull-request title
• I have verified that there are no overlapping pull-requests open
• I have verified that I am following matches the existing coding patterns and practice as demonstrated in the repository. These follow strict Stylecop rules 👮.
• I have provided test coverage for my change (where applicable)

Description

Apologies in advance, this touches a lot of files. Coverage is good though.

This is one of those breaking changes we unfortunately have to implement before V1

The current CommandParser was not injectable and also did not respect the current culture. Respecting culture is a required feature for using in many CMS implementations.

By default the system will use InvariantCulture so there are no surprises for people upgrading but there is a switch in the middleware options UseInvariantParsingCulture which allows using the CultureInfo.CurrentCulture.

Converters and Processors both required method signature changes to allow the update.

I also found and fixed two issues

1. We were disposing of the output stream outside of the write lock
2. We were not returning after a head request.

• Refactor converters, parser, and processors to allow current culture
• Update tests
• Add additional tests

[codecov]

Codecov Report

Merging #117 into master will increase coverage by 2.24%.
The diff coverage is 92.94%.

@@            Coverage Diff             @@
##           master     #117      +/-   ##
==========================================
+ Coverage   84.64%   86.89%   +2.24%     
==========================================
  Files          46       44       -2     
  Lines        1146     1190      +44     
  Branches      159      164       +5     
==========================================
+ Hits          970     1034      +64     
+ Misses        128      106      -22     
- Partials       48       50       +2     

Flag	Coverage Δ
#unittests	86.89% <92.94%> (+2.24%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/ImageSharp.Web/Commands/TypeConstants.cs	100.00% <ø> (ø)
...rc/ImageSharp.Web/Processors/FormatWebProcessor.cs	100.00% <ø> (ø)
...b/Commands/Converters/SimpleCommandConverter{T}.cs	75.00% <66.66%> (+3.57%)	⬆️
.../ImageSharp.Web/Middleware/ImageSharpMiddleware.cs	79.26% <69.56%> (+0.32%)	⬆️
...ageSharp.Web/Commands/Converters/ColorConverter.cs	86.66% <84.61%> (-1.80%)	⬇️
...harp.Web/Middleware/ImageSharpMiddlewareOptions.cs	78.00% <87.50%> (+14.36%)	⬆️
src/ImageSharp.Web/Commands/CommandParser.cs	95.65% <95.23%> (-2.63%)	⬇️
...Sharp.Web/Commands/Converters/ArrayConverter{T}.cs	100.00% <100.00%> (ø)
...mageSharp.Web/Commands/Converters/EnumConverter.cs	85.71% <100.00%> (+1.09%)	⬆️
.../Commands/Converters/IntegralNumberConverter{T}.cs	100.00% <100.00%> (ø)
... and 9 more

---

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 b4b3fd5...d0b2e10. Read the comment docs.

[tocsoft]

just the one comment on Enum handling but all the rest is looking good to me.

Only other thing that I have a small issue (mostly an aesthetics thing and not something I would explicitly want changing unless you wanted to) is the DefaultServicesAreRegistered tests... that to me is screaming out to be a theory test taking in the Service and implementation types as arguments.

[JimBobSquarePants]

@tocsoft I updated the parser to allow custom enum types also fixed some boxing allocations caused by the parser/converter API

Agree re the test but left it ugly as-is for now as I didn't want to end up potentially fighting Xunit skipping theory data values because it thinks they're all the same. Time is precious! 😄