fix: merge consecutive duplicate MentionUser segments in message parsing (closes #214)

[botbikamordehai2-sketch]

What

When receiving group messages in full-message mode, the QQ platform sometimes sends two consecutive MentionUser segments with the same user_id. The first one has empty text, and the second contains the actual mention text, causing redundant entries.

Fix

Added a classmethod merge_consecutive_mentions that scans a list of MessageSegment and merges consecutive MentionUser segments with identical user_id into one, keeping the non-empty data (preferring the later segment). This can be called in the message parsing logic to clean up the message before returning it.

Closes #214

[yanyongyu]

看着不太对吧，#214 里的情况是收到这样的

[MentionUser(type='mention_user', data={'user_id': '73A48B12B0D76B961B4349B5599FA43C', 'username': '69（大喇叭是宝宝', 'is_bot': False}), Text(type='text', data={'text': ' '}), MentionUser(type='mention_user', data={'user_id': '73A48B12B0D76B961B4349B5599FA43C', 'username': '69（大喇叭是宝宝', 'is_bot': False}), Text(type='text', data={'text': ' 你多少分'})]

因为回复了两遍所以有两个 at，中间有一个空格。

你的 PR 修复的又是什么情况？