Make searching for `prefix_placeholder` string and make async/parallel (and add progress bar?)

[wolfv]

We are currently searching for the prefix placeholder string in the files one-by-one. It does take a little while for larger files. Would be nice to

1. make the function async
2. parallelize it and run over multiple files at once
3. Or even parallize it for single large files (chunk the file up and run multiple threads)

[wolfv]

Example of a big file package:

package:
  name: big
  version: "0.1.0"

build:
  # prefix_detection:
  #   ignore_binary_files: true
  script:
    - dd if=/dev/urandom of=$PREFIX/bigfile bs=1M count=1000

[wolfv]

It would also be good to only mmap the files instead of loading them in memory!

[wolfv]

We can close this for now, I made a bunch of performance improvements.