Remove dependency on Apache Commons HttpClient 3.1

[ato]

HttpClient 3 was discontinued in 2007 and frequently triggers alerts in dependency vulnerability scanners. We're also not using much of it anymore, with one big exception.

The URI class is the foundation of UsableURI and central to Heritrix which has made removing the library difficult. URIException in particular appears a lot in client code. HttpClient 4+ has switched to java.net.URI and the main reason Heritrix was built on HttpClient URI instead was because java.net.URI is not flexible and differs from how browsers behave. (Although, how browsers behave has shifted over time.)

Eventually we'll probably need to rework Heritrix's URI handling to follow the WhatWG URL spec. However, to let us remove the dependency while keeping UsableURI working, this copies HttpClient 3's URI, URIException and ChunkedInputStream with some small tweaks remove their dependency on other classes in HttpClient. The HttpClient Header class is replaced with our existing HttpHeader. URI and ChunkedInputStream are marked package private for now.

This is a breaking API change and will require a major version bump. If you're using webarchive-commons, you'll need to make the following updates to your code:

1. Update URIException import

   Before:

   import org.apache.commons.httpclient.URIException;

   After:

   import org.archive.url.URIException;

2. Replace Header with HttpHeader

   Before:

   import org.apache.commons.httpclient.Header
   
   Header[] headers = LaxHttpParser.parseHeaders(stream, charset);

   After:

   import org.archive.format.http.HttpHeader
   
   HttpHeader[] headers = LaxHttpParser.parseHeaders(stream, charset);

Fixes #78