Remove unsafe code from the Uri.UnescapeString helper

[MihaZupan]

Closes #31173

We have an UnescapeString helper that escapes/unescapes input based on various flags (copy input as-is, unescape only, escape reserved chars, or both).

This PR removes the use of unsafe code from this helper.

I also removed the "copy input as-is" and "unescape only" modes from this helper. For copy-only we just forward to CopyTo/Append directly, whereas "unescape only" gets a dedicated helper since it's a reasonably hot path (UnescapeDataString calls into this).

Initial benchmarks look good (no meaningful difference): MihuBot/runtime-utils#1612

Alternating escaped/unescaped

Method	Toolchain	Mean	Error	Ratio
Unescape	main	214.6 ns	4.02 ns	1.00
Unescape	pr	197.5 ns	2.73 ns	0.92

Only unescaped

Method	Toolchain	Mean	Error	Ratio
Unescape	main	270.0 ns	0.79 ns	1.00
Unescape	pr	262.9 ns	2.34 ns	0.97

Escaped at start followed by unescaped

Method	Toolchain	Mean	Error	Ratio
Unescape	main	88.41 ns	0.356 ns	1.00
Unescape	pr	31.44 ns	0.228 ns	0.36

[Copilot]

Pull Request Overview

This PR refactors the URI unescaping logic by:

• Introducing a simpler Unescape method that unconditionally unescapes percent-encoded sequences
• Removing the UnescapeAll flag from UnescapeMode enum
• Renaming CopyOnly to None in UnescapeMode enum
• Removing unused unsafe code and the System.Runtime.InteropServices import
• Consolidating various unescape code paths to use the new Unescape method where full unescaping is desired

Key changes:

• New UriHelper.Unescape() method provides simple unconditional unescaping
• Callsites that previously used UnescapeMode.Unescape | UnescapeMode.UnescapeAll now use the simpler Unescape() method
• The more complex UnescapeString() method is retained for cases requiring conditional escaping/unescaping with reserved character handling

Reviewed Changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
UriHelper.cs	Adds new Unescape() method, refactors UnescapeString() to use span-based operations instead of pointers, removes unsafe code
UriExt.cs	Updates UnescapeDataString and TryUnescapeDataString to use new Unescape() method
UriEnumTypes.cs	Renames CopyOnly to None, removes UnescapeAll flag
Uri.cs	Updates multiple callsites to use new Unescape() method or simplified logic, removes pointer-based operations
PercentEncodingHelper.cs	Adds scoped modifier to parameter, minor cleanup

Comments suppressed due to low confidence (1)

src/libraries/System.Private.Uri/src/System/Uri.cs:1081

• Potential index out of bounds exception for UNC paths. This code assumes index 1 contains a drive letter separator, but for UNC paths, result[1] would be the second backslash character. This check should be conditional on IsDosPath or include bounds checking.

                if (result[1] == '|')
                    result[1] = ':';

[EgorBo]

Nice!

[MihaZupan]

/ba-g android timing out