Improve Performance Of Unicode Highlighting / Link Detection

# Problem Currently, both the unicode highlighting and link detection feature scan the entire text model again whenever a change has been made (after some timeout). This is done in a webworker and thus does not impact performance of the main UI. But it might affect overall system performance and energy consumption. # Relativation Nonetheless, even for huge files (8MB, more than 230k lines) link detection takes about 70ms and unicode highlighting about 20ms on my machine in the webworker. For files with hundreds of thousands of links however, updating the decorations in the UI takes up to a second. Also, sending the decoration becomes more expensive than computing them. Thus, I don't think this improvement is currently needed, but it might be interesting to explore this idea in the futer. # Proposed Solution We could introduce an mechanism to incrementally compute and update decorations. Such a solution could look like this: ```ts interface IMatch<T> { range: Range; data: T; } interface WhitespaceIndependentTextMatcher<T> { isWhitespaceIndependent: true; /** * ``` * ∀txt1 txt2: String, WS: (\r\n\t\u{20})* * findMatches(txt1 + WS + txt2) * = [ * ...findMatches(txt1), * ...findMatches(txt2).map(d => ({ ...d, range: d.range.moveBy(txt1 + WS) })) * ] * ``` */ findMatches(text: string): IMatch<T>[]; } interface IncrementalTextMatcher<T> { getMatches(): IMatch<T>; handleEdit(text: string, oldRange: Range): void; } function createIncrementalMatcher<T>(matcher: WhitespaceIndependentTextMatcher<T>, textModel: ILineBasedTextModel): IncrementalTextMatcher<T> { // ... } ```