Add regression coverage for InputRichMessage rich-text serialization shapes

[Copilot]

This addresses the review comment on rich message serialization coverage. It adds focused regression tests for the three supported RichText input shapes when serialized through InputRichMessage / InputRichBlock, including a subtype with required extra fields.

• Coverage added

  • plain string block text
  • nested list-based rich text
  • nested RichText objects

• Subtype regression guard

  • adds explicit coverage for RichTextDateTime nested inside another rich-text node
  • asserts unix_time and date_time_format survive serialization so incomplete subtype payloads are caught

• Scope

  • test-only change in tests/test_types.py
  • no production behavior changes

message = types.InputRichMessage(
    blocks=[
        types.InputRichBlockParagraph(
            text=types.RichTextBold(
                types.RichTextDateTime(
                    text='Mar 1',
                    unix_time=1740787200,
                    date_time_format='short',
                )
            )
        )
    ]
)

assert message.to_dict()['blocks'][0]['text'] == {
    'type': 'bold',
    'text': {
        'type': 'date_time',
        'text': 'Mar 1',
        'unix_time': 1740787200,
        'date_time_format': 'short',
    },
}

[Badiboy]

@coder2020official Do we need this? :)

[coder2020official]

I feel like it's useless, our tests coverage is non-existent anyways and this PR doesn't do much