Bug: Brotli cannot be detected based on magic bytes

[NodyHub]

The current implementation tries to detect brotli archives based on magic bytes:

A test has shown that the listed magic bytes cannot be used to detect every compressed archive:

~/tmp/goextract-test% date > test
~/tmp/goextract-test% brotli test
~/tmp/goextract-test% ll
total 16
-rw-r--r--  1 jan  staff    29B Mar 28 14:51 test
-rw-r--r--  1 jan  staff    34B Mar 28 14:51 test.br
~/tmp/goextract-test% rm test
test
~/tmp/goextract-test% goextract test.br
2024/03/28 14:51:32 error during extraction: archive type not supported
~/tmp/goextract-test% xxd test.br
00000000: 2170 0004 5468 7520 4d61 7220 3238 2031  !p..Thu Mar 28 1
00000010: 343a 3531 3a31 3420 4345 5420 3230 3234  4:51:14 CET 2024
00000020: 0a03

A further investigation revealed that the designer of brotli do not consider adding magic bytes to the algorithm:
google/brotli#298 (comment)

We need to adjust the brotli detection to be based in file extension.