Example in find_chunk_edge() can panic 'range start index'

[allada]

Using some example code here:
https://docs.rs/hash-roll/0.3.0/hash_roll/trait.Chunk.html#tymethod.find_chunk_edge

It turns out that discard_ct could result in data.len() + 1 in size as seen in this code:
https://docs.rs/hash-roll/0.3.0/src/hash_roll/fastcdc.rs.html#75

The only happens when the boundary is on the last byte and can result in i equaling data.len() causing &data[..data.len() + 1] resulting in a panic.

The fix is simple, I might suggest adding something like this to the docs:

if discard_ct > data_buf.len() {
    break;
}

Thanks for the hard work in providing this library!

[safinaskar]

@jmesmon , example in https://docs.rs/hash-roll/0.3.0/hash_roll/trait.Chunk.html#tymethod.find_chunk_edge seems to be absolutely incorrect. It initializes prev_cut to 0 and then never sets it again. So it seems something is really wrong here. So I'm afraid to use this library, because there is no example I can rely on

[safinaskar]

Also, it seems that &orig_data[prev_cut..g_cut] instead of &orig_data[prev_cut..cut_point] should be. There may be other bugs in this code, so, please, proof read it