Improve support for import of "lots of data"

[gggeek]

Example things that could help in massive-content-creation scenarios:

• switch off db transactions
• disable calls to Solr indexation
• disable emitting events for each loop step (to be verified)
• disable sf debug (of course)
• when downloading images off the web, cache them locally for reuse to avoid going over quotas
• support for multiple image providers to avoid going over quotas
• other ?

[gggeek]

"when downloading images off the web, cache them locally for reuse to avoid going over quotas" => implemented in release 1.4