Missing glyphs and inccorect breaks when setting MaxLines on TextBlock containing line breaks

[handerss-spotfire]

Setting the MaxLines property on a TextBlock that contains text with line breaks results in missing glyphs being rendered at the line break character positions.

Repro:

var tb = new TextBlock();
tb.AddText("Hello\n\nWorld!", new Style());
tb.MaxLines = 1;

Result:

[handerss-spotfire]

Maybe it enough to always line break at a line break character if that occurs before the max width of the text block?

E.g:

diff --git a/Topten.RichTextKit/TextBlock.cs b/Topten.RichTextKit/TextBlock.cs
index 1b29bda..9403b80 100644
--- a/Topten.RichTextKit/TextBlock.cs
+++ b/Topten.RichTextKit/TextBlock.cs
@@ -1475,10 +1475,15 @@ namespace Topten.RichTextKit
                     fr = _fontRuns[frIndex];
                     var room = _maxWidthResolved - fr.XCoord;
                     frSplitIndex = frIndex;
-                    codePointIndexSplit = fr.FindBreakPosition(room, frSplitIndex == frIndexStartOfLine);
-                    codePointIndexWrap = codePointIndexSplit;
-                    while (codePointIndexWrap < _codePoints.Length && UnicodeClasses.LineBreakClass(_codePoints[codePointIndexWrap]) == LineBreakClass.SP)
-                        codePointIndexWrap++;
+                    var breakPosition = fr.FindBreakPosition(room, frSplitIndex == frIndexStartOfLine);
+
+                    if (breakPosition < codePointIndexSplit)
+                    {
+                        codePointIndexSplit = breakPosition;
+                        codePointIndexWrap = codePointIndexSplit;
+                        while (codePointIndexWrap < _codePoints.Length && UnicodeClasses.LineBreakClass(_codePoints[codePointIndexWrap]) == LineBreakClass.SP)
+                            codePointIndexWrap++;
+                    }
                 }
 
                 // Split it

[handerss-spotfire]

The patch above is a bit too naive since codePointIndexSplit can be negative if no non-character breaks were found. I have modified it for our use-case such that it differentiates between required text breaks and others, and that it prefers to break at characters if that happens later than a break position (leads to behavior similar to StringTrimming.EllipsisCharacter in GDI+).

I'll submit a PR if testing is successful.