setDelimiterSpecs() accumulates markLength instead of recalculating

In `src/main/java/org/apache/maven/shared/filtering/MultiDelimiterInterpolatorFilterReaderLineEnding.java`, lines 133–141:

```
public AbstractFilterReaderLineEnding setDelimiterSpecs(Set specs) {
delimiters.clear();
for (String spec : specs) {
delimiters.add(DelimiterSpecification.parse(spec));
markLength += spec.length() * 2;
}
return this;
}
```

`markLength` is accumulated on top of the current value rather than recalculated from base (255 + escapeString.length). If `setDelimiterSpecs` is called multiple times, `markLength` grows unbounded. Currently this is safe only because `setEscapeString()` always follows in `Wrapper.getReader()` and calls `calculateMarkLength()` which resets the value. However, this inter-call dependency is fragile and undocumented.

Start in src/main/java/org/apache/maven/shared/filtering/MultiDelimiterInterpolatorFilterReaderLineEnding.java, focusing on setDelimiterSpecs() and calculateMarkLength(). Trace how markLength is initialized and updated, then verify that calling setDelimiterSpecs() repeatedly recalculates from the base value rather than accumulating. Run the relevant Maven test suite and add regression coverage if the existing tests do not exercise repeated calls.

Written by the indexing model from the issue text.