Fix HMM segmentation of digit+letter tokens to align with Python jieba

[Copilot]

C++ cppjieba splits alphanumeric combinations like "5G" and "3D" into separate tokens, while Python jieba's finalseg keeps them together via re_skip = [a-zA-Z0-9]+(?:\.\d+)?.

Root cause

HMMSegment::NumbersRule consumed only [0-9.]+, so "5G" produced ["5", "G"] instead of ["5G"].

Changes

• HMMSegment::NumbersRule: Inner loop changed from [0-9.] to [a-zA-Z0-9], allowing letters after an initial digit. Replaced multi-dot permissiveness with a single optional decimal suffix \.\d+, matching Python's pattern.
• HMMSegment::SequentialLetterRule: Added the same optional \.\d+ suffix (e.g. abc1.2 stays together).

Before:  "3D打印" → ["3", "D", "打印"]
After:   "3D打印" → ["3D", "打印"]   ✓ matches Python jieba

Before:  "5G网络" → ["5", "G", "网络"]
After:   "5G网络" → ["5G", "网络"]   ✓ matches Python jieba

         "3.5KG"  → ["3.5", "KG"]    ✓ unchanged (decimal stops letter merging)

• Tests: Added HMMSegmentTest.AlphanumericCombinations and two new cases in MixSegmentTest.Test1 covering the above patterns.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>c++与python版本切词不对齐</issue_title>
<issue_description>用c++切词和python切词结果边界和词性不完全一致，请问有能对齐的版本吗？</issue_description>

Comments on the Issue (you are @copilot in this section)

@yanyiwu 目前还没有完全对齐的。毕竟两个仓库都在变化。完全一致比较难。只不过差异程度应该不大才对。 @yanyiwu 为避免分散讨论，我会把 #197 和 #198 归并到本 issue，统一跟踪 C++/Python 分词边界不一致（尤其是数字+单位、字母数字组合）问题。后续如果有可复现样例，请优先补充到这里：\n\n- 输入原文（UTF-8）\n- C++ 输出\n- Python jieba 输出\n- 使用词典与版本信息\n\n这样便于集中评估规则差异与可行修复。

• Fixes c++与python版本切词不对齐 #192

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.