CSV/TSV use a pool of buffered readers to avoid allocs

[gabriel-vasile]

use a pool of buffers to alleviate memory allocs in csv; related to #553

When iterating over multiple files, csv detector allocated a new buffer
for each file. This change adds a pool of buffers that can be reused
between detections. The same pool is shared between csv and tsv
detectors.

Benchmark results:

➜  magic git:(csvpprof) ✗ benchstat before after
goos: linux
goarch: amd64
pkg: github.com/gabriel-vasile/mimetype/internal/magic
cpu: Intel(R) Core(TM) i7-10510U CPU @ 1.80GHz
      │   before    │               after                │
      │   sec/op    │   sec/op     vs base               │
Csv-8   3.052µ ± 3%   1.695µ ± 3%  -44.46% (p=0.001 n=7)

      │    before    │                after                │
      │     B/op     │     B/op      vs base               │
Csv-8   5.258Ki ± 0%   1.164Ki ± 0%  -77.86% (p=0.001 n=7)

      │   before   │               after               │
      │ allocs/op  │ allocs/op   vs base               │
Csv-8   13.00 ± 0%   11.00 ± 0%  -15.38% (p=0.001 n=7)