Optimize UTC-offset adjustment in Calendar.ISO parsing

[preciz]

Assisted-by: Codex:GPT-6

~1.3x faster and uses less memory.

Normalize nonzero UTC offsets directly in seconds and reuse seconds_to_time/1, avoiding intermediate day-fraction conversions while preserving parsed microseconds and precision.

Bench:

# Run with: elixir bench.exs
Mix.install([:benchee])

path = "lib/elixir/lib/calendar/iso.ex"

for {ref, module} <- [{"main", Old}, {"HEAD", New}] do
  {source, 0} = System.cmd("git", ["show", "#{ref}:#{path}"], cd: __DIR__)

  source
  |> String.replace("defmodule Calendar.ISO do", "defmodule #{inspect(module)} do")
  |> Code.compile_string(path)
end

Benchee.run(
  %{"main" => &Old.parse_utc_datetime/1, "HEAD" => &New.parse_utc_datetime/1},
  inputs: %{
    "positive, same day" => "2024-03-01T12:34:56.123456+02:30",
    "positive, previous day" => "2024-03-01T00:15:23.123456+02:30",
    "negative, same day" => "2024-03-01T12:34:56.123456-02:30",
    "negative, next day" => "2024-12-31T23:45:01.123456-02:30"
  },
  pre_check: :all_same,
  warmup: 1,
  time: 3,
  memory_time: 1
)

Results:

Operating System: Linux
CPU Information: AMD Ryzen 7 8845HS w
Number of Available Cores: 16
Available memory: 54.72 GB
Elixir 1.21.0-dev
Erlang 29.0.5
JIT enabled: true

Benchmark suite executing with the following configuration:
warmup: 1 s
time: 3 s
memory time: 1 s
reduction time: 0 ns
parallel: 1
inputs: negative, next day, negative, same day, positive, previous day, positive, same day
Estimated total run time: 40 s
Excluding outliers: false

##### With input negative, next day #####
Name           ips        average  deviation         median         99th %
HEAD        3.90 M      256.23 ns  ±2078.02%         230 ns         371 ns
main        3.12 M      320.46 ns  ±1322.08%         291 ns         441 ns

Comparison: 
HEAD        3.90 M
main        3.12 M - 1.25x slower +64.24 ns

Memory usage statistics:

Name    Memory usage
HEAD           616 B
main           792 B - 1.29x memory usage +176 B

**All measurements for memory usage were the same**

##### With input negative, same day #####
Name           ips        average  deviation         median         99th %
HEAD        4.54 M      220.15 ns  ±2260.04%         191 ns         331 ns
main        3.29 M      303.67 ns  ±1560.26%         271 ns         421 ns

Comparison: 
HEAD        4.54 M
main        3.29 M - 1.38x slower +83.52 ns

Memory usage statistics:

Name    Memory usage
HEAD           616 B
main           792 B - 1.29x memory usage +176 B

**All measurements for memory usage were the same**

##### With input positive, previous day #####
Name           ips        average  deviation         median         99th %
HEAD        3.96 M      252.36 ns  ±2284.39%         230 ns         370 ns
main        3.02 M      331.67 ns  ±1502.95%         301 ns         451 ns

Comparison: 
HEAD        3.96 M
main        3.02 M - 1.31x slower +79.31 ns

Memory usage statistics:

Name    Memory usage
HEAD           616 B
main           792 B - 1.29x memory usage +176 B

**All measurements for memory usage were the same**

##### With input positive, same day #####
Name           ips        average  deviation         median         99th %
HEAD        4.68 M      213.69 ns  ±2346.88%         190 ns         321 ns
main        3.17 M      315.05 ns  ±1518.93%         281 ns         441 ns

Comparison: 
HEAD        4.68 M
main        3.17 M - 1.47x slower +101.36 ns

Memory usage statistics:

Name    Memory usage
HEAD           616 B
main           792 B - 1.29x memory usage +176 B

**All measurements for memory usage were the same**

[josevalim]

💚 💙 💜 💛 ❤️