Add UTF-8 APIs on TextEncoder and add JavascriptEncoder to JsonEncodedText

[steveharter]

Replacement for PR #38947 which was started by @ahsonkhan

This is the first phase of https://github.com/dotnet/corefx/issues/37192; additional phases will allow the custom encoders to be specified in writer and serializer options.

• Update tests to have them succeed (match semantics of JavaScriptEncoder) and add new tests.
• Change encoding of forward slash from "u002f" to "/" to match current semantics of JavaScriptEncoder. Note this will likely change soon to not escape at all per @GrabYourPitchforks.
• Improved perf of ascii scenarios in JavaScriptEncoder from being 4x slower to ~25% slower (1.25x) than the implementation in System.Text.Json.
• In order to prevent a regression of ~25% by default for ascii (when not specifying a custom encoder) the current code in System.Text.Json was kept. In addition, the code was modified to improve custom encoding scenarios by ~15%. If a non-ascii (latin) chacter >= 128 is used, then the default JavaScriptEncoder is used.

[steveharter]

/azp run corefx-ci

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[ericstj]

Build failures are occurring because system.text.json is inbox on uap but text.encoding.web is not.

corefx/src/System.Text.Encodings.Web/Directory.Build.props

Line 6 in be24849

<IsNETCoreApp>true</IsNETCoreApp>

To fix this, make s.t.e.w inbox on uap. See example 10fc6ee

[GrabYourPitchforks]

The change to DecodeScalarValueFromUtf8 is the only thing that unsettled me, everything else is primarily just a nit.

[GrabYourPitchforks]

Per the offline conversation, there was a desire to change '/' to go through unescaped by default. Feel free to make this change if you wish, otherwise it will eventually be addressed per https://github.com/dotnet/corefx/issues/39445.

[steveharter]

I thought it would be good to handle that in a separate PR.