add node depth limit

[spirit-at-canva]

Why

I want to limit the max depth of node traversal.

What

Apply an optional limit to the depth of node parsing.

We make a conscious design choice to drop nodes that are too deep, instead of failing fast.

We apply the limit when constructing the AST, by dropping deeper node generation (we collapse it into plain text).

When visiting the AST, we apply a depth limit again and drop deep nodes altogether.

The two depths correspond to different meaning, and aren't semantically the same.

[spirit-at-canva]

cc @robinst

[robinst]

Changes to DocumentParser, Parser, ParserTest LGTM, with some minor comments to address.

The visitor changes: Can you take them out of this PR? I'm not sure I want to include this by default on every visitor, and it's something that's pretty straightforward to add in subclasses (overriding the protected visitChildren method).

[robinst]

The top-level block parser is really an implementation detail, no need to mention it prominently:

Suggested change

	* Limit how many non-document block parsers may be open at once while parsing.
	* Limit how many block parsers may be open at once while parsing.

[robinst]

Let's use a MarkdownRenderer instead (TextContentRenderer is an arbitrary plain text format, we don't want to rely on it for tests):

Suggested change

	return TextContentRenderer.builder().build().render(node).trim();
	return MarkdownRenderer.builder().build().render(node).trim();

[robinst]

Incrementing and decrementing should probably be done outside the while loop, just once?