Fix/unicode surrogate pairs

[springcomp]

Fixes #74.

Internally, the .NET String class uses UTF-16 which encodes characters from the Basic Multilingual Plane.
Characters outside this plan are encoded as a sequence of two UTF-16 code units called a surrogate pair.
In that case, a single Unicode character – identified by its given codepoint – is encoded as two UTF-16 code units.
In that case, the .NET String length property returns 2 even though there is a single character.

Another specials case is composite characters. For instance, the character é can be encoded as two different Unicode sequences:

• U+00E9 LATIN SMALL LETTER E WITH ACUTE i.e a single Unicode codepoint.
• U+0065 LATIN SMALL LETTER E, U+0301 COMBINING ACUTE ACCENT i.e a sequence of two Unicode codepoints.

As a result, the builtin functions length(), reverse(), sort() and sort_by() did not handle those cases correctly.

This PR fixes #74 by introducing proper support for strings as a sequence of codepoints.
However, this still raises some issues with regards to composite characters.

[springcomp]

For the record, this PR fixes surrogate pairs.
However all other processing operations happen on the codepoint level.