feat/Support custom extensions for partition_md

[Chenyl-Sai]

Is your feature request related to a problem? Please describe.
In unstructured/partition/md.py, the partition_md method converts Markdown to HTML using a fixed extension list: html = markdown.markdown(text, extensions=["tables"]). This causes an issue when processing code blocks containing # comments, as they get incorrectly parsed as <h1> tags.

Describe the solution you'd like
Modify partition_md to accept custom Markdown extensions via kwargs:

1. Read extensions parameter from method kwargs
2. Default to ["tables"] if not specified (backward compatible)
3. Pass extensions to markdown.markdown()
4. Allows users to handle special cases (e.g. add "fenced_code" for code blocks)

Describe alternatives you've considered
Security Impact: None (parameter addition only)
Backward Compatibility:

• Low risk: Existing calls without extensions kwarg remain unchanged
• Medium risk: If users already pass extensions in kwargs (unlikely given current usage)

Recommended Implementation Approach:

1. Defensive Kwargs Handling:

   def partition_md(
    filename: str | None = None,
    file: IO[bytes] | None = None,
    text: str | None = None,
    url: str | None = None,
    metadata_filename: str | None = None,
    metadata_last_modified: str | None = None,
    **kwargs: Any,) -> list[Element]:
       ....
    extensions = kwargs.pop('extensions', ["tables"]) 
    if not (isinstance(extensions, list) and all(isinstance(ext, str) for ext in extensions)):
        logging.warning(
            f"Ignoring invalid 'extensions' argument (expected list of strings): {extensions!r}"
        )
        extensions = ["tables"]
    html = markdown.markdown(text, extensions=extensions)
       ....

Additional context
The following is the test markdown text

# create the container
docker run -dt --name unstructured downloads.unstructured.io/unstructured-io/unstructured:latest

And the test code before modified

from unstructured.partition.md import partition_md
elements = partition_md(text=text)
for i, el in enumerate(elements):
    print(f"\n--- Element {i} ---")
    print(f"Type: {type(el).__name__}")
    print(f"Category: {getattr(el, 'category', 'N/A')}")
    print(f"Text: {el.text!r}") 

And the outputs:

--- Element 0 ---
Type: Text
Category: UncategorizedText
Text: '```bash'

--- Element 1 ---
Type: Title
Category: Title
Text: 'create the container'

--- Element 2 ---
Type: Text
Category: UncategorizedText
Text: 'docker run -dt --name unstructured downloads.unstructured.io/unstructured-io/unstructured:latest ```'

Here is the test code after modified

from unstructured.partition.md import partition_md
elements = partition_md(text=text, **{'extensions': ['fenced_code']})
for i, el in enumerate(elements):
    print(f"\n--- Element {i} ---")
    print(f"Type: {type(el).__name__}")
    print(f"Category: {getattr(el, 'category', 'N/A')}")
    print(f"Text: {el.text!r}") 

And the outputs:

Type: NarrativeText
Category: NarrativeText
Text: '# create the container\ndocker run -dt --name unstructured downloads.unstructured.io/unstructured-io/unstructured:latest'