PERF: Frequency determination when concatting datetimes

[rhshadrach]

Ref: pandas-dev/asv-runner#125

Addresses the bottom 4 benchmarks there.

import pandas as pd

idx = pd.date_range("2024-01-01", periods=30 * 24 * 3600, freq="s")
chunks = [
    pd.Series(range(len(c)), index=c) 
    for c in (idx[i:i + 2592] for i in range(0, len(idx), 2592))
]
%timeit pd.concat(chunks)
# 23.4 ms ± 186 μs per loop (mean ± std. dev. of 7 runs, 10 loops each)   <--- main
# 16.8 ms ± 119 μs per loop (mean ± std. dev. of 7 runs, 100 loops each)  <--- PR

[jbrockmendel]

do we know these all have the same unit?