[Feature] On-the-fly checksum calculation

[cnsgithub]

I've got the requirement to avoid duplicate downloads when searching for special file types. This can be achieved by simply comparing checksums. Since it's not security-relevant, MD5 hashes should do.

Since in crawler4j each file is represented by a Page and consumed anyways in Page.toByteArray, it would be an advantage in performance to calculate it there on the fly by simply decorating the InputStream with DigestInputStream, instead of touching the byte array again afterwards.

@yasserg What do you think? Do you want me to provide a PR? Would you prefer to make this feature an opt-in?

[cnsgithub]

PR: #400