Crash on unicode characters used after string interpolation

sourcekitten structure returns wrong key.length when unicode characters are used after string interpolation. Here is an example:

sourcekitten structure --text 'class Foo { var blah: String { switch { default: return "\(true) π" }}}'

It will always give 26 for length of case statement:

"key.kind" : "source.lang.swift.stmt.case",
"key.offset" : 40,
"key.length" : 26

At the same time when unicode character comes before string interpolation length is completely different.

sourcekitten structure --text 'class Foo { var blah: String { switch { default: return "π \(true)" }}}'

"key.kind" : "source.lang.swift.stmt.case",
"key.offset" : 40,
"key.length" : 32

but again adding whitespaces after string interpolation does not change the length, which will probably lead to crash again.

Crash happens when trying to extract string for such statement from the source:

frame #7: 0x000000010e443ac0 SourceKittenFramework`NSString.CacheContainer.location(byteOffset=91, self=0x00007fabf73072d0) at String+SourceKitten.swift:129

Reference
https://github.com/krzysztofzablocki/Sourcery/issues/432

Start with the sourcekitten structure --text reproductions and inspect String+SourceKitten.swift around NSString.CacheContainer.location at line 129. Compare the reported offsets and lengths for unicode before and after interpolation, then verify that extracting the case statement no longer crashes and that the reported length is consistent.