Change loop in parse_attributes to while true.

[naitoh]

Why

loop is slower than while true.

Benchmark

RUBYLIB= BUNDLER_ORIG_RUBYLIB= /Users/naitoh/.rbenv/versions/3.3.0/bin/ruby -v -S benchmark-driver /Users/naitoh/ghq/github.com/naitoh/rexml/benchmark/parse.yaml
ruby 3.3.0 (2023-12-25 revision 5124f9ac75) [arm64-darwin22]
Calculating -------------------------------------
                         before       after  before(YJIT)  after(YJIT)
                 dom     11.186      11.304        17.395       17.450 i/s -     100.000 times in 8.940144s 8.846590s 5.748718s 5.730793s
                 sax     30.811      31.629        47.352       48.040 i/s -     100.000 times in 3.245601s 3.161619s 2.111854s 2.081594s
                pull     35.793      36.621        56.924       57.313 i/s -     100.000 times in 2.793829s 2.730693s 1.756732s 1.744812s
              stream     33.157      34.757        46.792       50.536 i/s -     100.000 times in 3.015940s 2.877088s 2.137106s 1.978787s

Comparison:
                              dom
         after(YJIT):        17.4 i/s
        before(YJIT):        17.4 i/s - 1.00x  slower
               after:        11.3 i/s - 1.54x  slower
              before:        11.2 i/s - 1.56x  slower

                              sax
         after(YJIT):        48.0 i/s
        before(YJIT):        47.4 i/s - 1.01x  slower
               after:        31.6 i/s - 1.52x  slower
              before:        30.8 i/s - 1.56x  slower

                             pull
         after(YJIT):        57.3 i/s
        before(YJIT):        56.9 i/s - 1.01x  slower
               after:        36.6 i/s - 1.57x  slower
              before:        35.8 i/s - 1.60x  slower

                           stream
         after(YJIT):        50.5 i/s
        before(YJIT):        46.8 i/s - 1.08x  slower
               after:        34.8 i/s - 1.45x  slower
              before:        33.2 i/s - 1.52x  slower

• YJIT=ON : 1.00x - 1.08x faster
• YJIT=OFF : 1.01x - 1.04x faster

[kou]

I merged this but I don't want to do this optimization as much as possible personally...