[easy] minor doc update for --rsyncable

[Cyan4973]

rephrase the speed impact, as suggested by @devZer0.
fix #3567

[devZer0]

There is a negligible impact to compression ratio, and a small impact to compression speed,
only perceptible at higher speeds, typically when combining --rsyncable with many parallel
worker threads.

thank you - but - sorry - i don't want to be dogmatic, nor do i want to be pedantic , rude or a nagger - but even in this optimzed form, what's being claimed here simply isn't true.

otherwise, my perception of "negligible impact" is at least very different from your's.

let me show the difference again:

# dd if=/dev/zero of=zeroes.dat bs=1024k count=4096
4096+0 records in
4096+0 records out
4294967296 bytes (4.3 GB, 4.0 GiB) copied, 2.66998 s, 1.6 GB/s

# dd if=/dev/urandom of=random.dat bs=1024k count=4096
4096+0 records in
4096+0 records out
4294967296 bytes (4.3 GB, 4.0 GiB) copied, 20.9213 s, 205 MB/s

# time /root/zstd/zstd -c random.dat >/dev/null

real	0m4.699s
user	0m7.190s
sys	0m2.999s

# time /root/zstd/zstd -c zeroes.dat >/dev/null

real	0m2.433s
user	0m3.518s
sys	0m2.477s

# time /root/zstd/zstd -c random.dat --rsyncable >/dev/null

real	0m19.598s
user	0m25.092s
sys	0m4.188s

# time /root/zstd/zstd -c zeroes.dat --rsyncable >/dev/null

real	0m18.160s
user	0m20.175s
sys	0m3.692s

[Cyan4973]

my perception of "negligible impact" is at least very different from your's.

The expression "negligible impact" is associated to the concept of "compression ratio", as in the sentence :

There is a negligible impact to compression ratio

If you were to report the compression ratio of the 2 corner case scenarios used as illustration, you would see that this statement holds up pretty well.

[devZer0]

you are right, i mixed that up. sorry for that.

and thank you for fixing the docs!