Importer limit

[nicolasfranck]

What I changed:

• make the batch size configurable by adding the parameter "limit". The old hidden parameter _max_results is therefore replaced by limit. A batch size of 10 was too small anyway.

• add parameter "total" that limits the total returned number of records.

• the extraction of tag name should not use the prefixed tag names in the hash response "meta". e.g. http://sru.bibsys.no/search/biblio?version=1.1&operation=searchRetrieve&startRecord=1&maximumRecords=0&recordSchema=dc&query=video

  in this case you get "srw:numberOfRecords" in the hash "meta" instead of "numberOfRecords".

• document parameters "limit" and "total".

• override method count from Catmandu::Iterable. This makes it easy to inspect an SRU repo without fully downloading all matching records.

[coveralls]

Coverage decreased (-5.5%) to 84.492% when pulling 9e2f16758383ad16cd4a3a788c59f1e4be55da52 on importer_limit into 653cd64796e2707de71d102d207d9d3d47a451e6 on dev.

[nicolasfranck]

The actual output of extraResponseData in the hash "meta" depended on the input XML: Before:

"srw:numberOfRecords" when shown as such in the XML.

Now:

numberOfRecords (always)

Other, non srw keys, preserve their prefixes.