cap nesting depth in MarkdownSlurper to avoid stack overflow

[netliomax25-code]

1. MarkdownSlurper walks the parsed CommonMark tree recursively (blocksToList/nodeToMap, and textOf/appendText) with one stack frame per nesting level and no depth bound.
2. CommonMark builds that tree iteratively, so a small but deeply nested document (a single ~8 KB line of nested block quotes, or a deeply nested list) overflows the walk and raises a StackOverflowError inside parse()/parseText(). That raw Error escapes the documented MarkdownRuntimeException that callers guarding untrusted input catch.
3. Added a configurable cap (maxNestingDepth, default 1000, property groovy.markdown.maxNestingDepth) checked once before the walk, throwing MarkdownRuntimeException past the limit, matching the nesting cap JsonSlurper already applies.

Validation: a 5000-deep block-quote input now throws MarkdownRuntimeException instead of overflowing, documents within the limit parse unchanged, and the added regression tests plus the existing MarkdownSlurper suite pass.