[66_5] Perf tuning on tm_reader::decode via index_of

[da-liii]

Why you open this Pull Request?

as title describe

What work have you done in the current Pull Request?

For 5m tm doc: 564ms -> 543ms

[jingkaimori]

further optimization can be done, and qtest has facilities for benchmark:

https://doc.qt.io/qt-6/qttestlib-tutorial5-example.html

[da-liii]

and qtest has facilities for benchmark:

we do not need micro bench here.

[da-liii]

Tested on Linux using sicp.tm (3.25M), there is no significant performance improvements.

[jingkaimori]

Performance is slightly improved:

Before:

PASS   : TestConverter::test_texmacs_to_tree_bench(29_1_1.tm)
RESULT : TestConverter::test_texmacs_to_tree_bench():"29_1_1.tm":
     0.018 msecs per iteration (total: 76, iterations: 4096)
PASS   : TestConverter::test_texmacs_to_tree_bench(46_3.tm)
RESULT : TestConverter::test_texmacs_to_tree_bench():"46_3.tm":
     0.394 msecs per iteration (total: 101, iterations: 256)
PASS   : TestConverter::test_texmacs_to_tree_bench(64_1.tm)
RESULT : TestConverter::test_texmacs_to_tree_bench():"64_1.tm":
     1.5 msecs per iteration (total: 98, iterations: 64)

After:

PASS   : TestConverter::test_texmacs_to_tree_bench(29_1_1.tm)
RESULT : TestConverter::test_texmacs_to_tree_bench():"29_1_1.tm":
     0.016 msecs per iteration (total: 69, iterations: 4096)
PASS   : TestConverter::test_texmacs_to_tree_bench(46_3.tm)
RESULT : TestConverter::test_texmacs_to_tree_bench():"46_3.tm":
     0.37 msecs per iteration (total: 95, iterations: 256)
PASS   : TestConverter::test_texmacs_to_tree_bench(64_1.tm)
RESULT : TestConverter::test_texmacs_to_tree_bench():"64_1.tm":
     1.4 msecs per iteration (total: 92, iterations: 64)

[da-liii]

剩下的单元测试和注释，我来加一下吧

[jingkaimori]

LGTM

[jingkaimori]

Modification should not be make on parser of tm. tmu parser needs these optimization.