Improve invisible char data encoding

I noticed that the code points [here](https://github.com/microsoft/vscode/blob/79c4f9b75db3b4f50468fc022cacf8ba6d640aff/src/vs/base/common/strings.ts#L1193-L1196) are often consecutive e.g. <img width="909" alt="image" src="https://user-images.githubusercontent.com/5047891/210737564-ba1b7664-4db0-4624-ad6d-f448001fd062.png"> Here's a suggestion for encoding them in an array: ``` [9,10,11,12,13,32,127,160,173,847,1564,4447,4448,6068,6069,6155,6156,6157,6158,7355,7356,8192,8193,8194,8195,8196,8197,8198,8199,8200,8201,8202,8203,8204,8205,8206,8207] ``` could be encoded as: ``` [9,4,32,127,160,173,847,1564,4447,1,6068,1,6155,3,7355,1,8192,15] ``` The array can be decoded in linear time by comparing an element to the element before it. If the element is smaller, then this indicates a range e.g. * 9 > 4 ==> range from 9 to 13 * 32 < 127 ==> single element 32 * 127 < 160 ==> single element 127 * 160 < 173 ==> single element 160 * 173 < 847 ==> single element 173 * 847 < 1564 ==> single element 847 * 1564 < 4447 ==> single element 1564 * 4447 > 1 ==> range from 4447 to 4448 * 6068 > 1 ==> range from 6068 to 6069 * 6155 > 3 ==> range from 6155 to 6158 * 7355 > 1 ==> range from 7355 to 7356 * 8192 > 15 ==> range from 8192 to 8207