parser_tsv: Improve parse performance and reduce memory allocation

[Watson1978]

Which issue(s) this PR fixes:
Fixes #

What this PR does / why we need it:
This commit replaces Hash[@keys.zip(values)] with a while loop in TSVParser#parse.

The previous implementation using Array#zip created multiple short-lived intermediate arrays per line, which caused significant GC pressure when processing a large amount of logs.
By using a simple while loop and direct hash assignment, we can completely eliminate these intermediate object allocations.

Furthermore, because Array#zip is implemented in C natively, it leaves little room for Ruby's JIT compilers.

Micro benchmark

require 'bundler/inline'
gemfile do
  source 'https://rubygems.org'
  gem 'benchmark-ips'
  gem 'benchmark-memory'
end

@values = Array.new(9) { |i| i.to_s * 20 }
@keys = ("k1".."k9").to_a

def benchmarks(x)
  x.report('zip') do
    Hash[@keys.zip(@values)]
  end

  x.report('each_with_index') do
    r = {}
    @keys.each_with_index do |k, i|
      r[k] = @values[i]
    end
  end

  x.report('while') do
    r = {}
    i = 0
    len = @keys.length
    while i < len
      r[@keys[i]] = @values[i]
      i += 1
    end
  end
end

Benchmark.ips do |x|
  benchmarks(x)
  x.compare!
end

Benchmark.memory do |x|
  benchmarks(x)
end

• Result (no JIT)

$ ruby zip.rb
ruby 4.0.3 (2026-04-21 revision 85ddef263a) +PRISM [x86_64-linux]
Warming up --------------------------------------
                 zip   130.961k i/100ms
     each_with_index   121.090k i/100ms
               while   163.611k i/100ms
Calculating -------------------------------------
                 zip      1.315M (± 0.2%) i/s  (760.72 ns/i) -      6.679M in   5.080866s
     each_with_index      1.250M (± 0.6%) i/s  (799.97 ns/i) -      6.297M in   5.037284s
               while      1.655M (± 0.4%) i/s  (604.23 ns/i) -      8.344M in   5.041905s

Comparison:
               while:  1654988.8 i/s
                 zip:  1314549.8 i/s - 1.26x  slower
     each_with_index:  1250053.2 i/s - 1.32x  slower

Calculating -------------------------------------
                 zip     1.384k memsize (     0.000  retained)
                        19.000  objects (     0.000  retained)
                         7.000  strings (     0.000  retained)
     each_with_index   824.000  memsize (     0.000  retained)
                         8.000  objects (     0.000  retained)
                         7.000  strings (     0.000  retained)
               while   824.000  memsize (   824.000  retained)
                         8.000  objects (     8.000  retained)
                         7.000  strings (     7.000  retained)

• Result (JIT)

$ ruby --jit zip.rb
ruby 4.0.3 (2026-04-21 revision 85ddef263a) +YJIT +PRISM [x86_64-linux]
Warming up --------------------------------------
                 zip   132.431k i/100ms
     each_with_index   144.321k i/100ms
               while   222.340k i/100ms
Calculating -------------------------------------
                 zip      1.247M (± 1.8%) i/s  (801.86 ns/i) -      6.357M in   5.098935s
     each_with_index      1.502M (± 0.8%) i/s  (665.56 ns/i) -      7.649M in   5.091197s
               while      2.272M (± 0.9%) i/s  (440.10 ns/i) -     11.562M in   5.088800s

Comparison:
               while:  2272192.3 i/s
     each_with_index:  1502491.1 i/s - 1.51x  slower
                 zip:  1247100.3 i/s - 1.82x  slower

Calculating -------------------------------------
                 zip     1.384k memsize (     0.000  retained)                                                                                                     19.000  objects (     0.000  retained)
                         7.000  strings (     0.000  retained)
     each_with_index   824.000  memsize (     0.000  retained)
                         8.000  objects (     0.000  retained)
                         7.000  strings (     0.000  retained)
               while   824.000  memsize (     0.000  retained)
                         8.000  objects (     0.000  retained)
                         7.000  strings (     0.000  retained)

Docs Changes:
N/A

Release Note:

• parser_tsv: improve parse performance and reduce memory allocation

[kenhys]

LGTM.

[Watson1978]

Thanks