register global blaze loader in run_algorithm

The blaze loader exists to make it easy to register new pipeline datasets in zipline by reflecting information from various sources. We can automatically convert a tabular blaze expression (in a given format) to a Pipeline expression and register this with a pipeline loader (see zipline.pipeline.loaders.blaze.from_blaze; however, this loader is not registered by default so you cannot use these datasets in the cli or run_algorithm. I propose we register the zipline.pipeline.loaders.blaze.global_loader by default so that these can be used without major modification to zipline. If you want to register custom sub-loaders, you are an expert so this is not intended to help your workflow.

Does anyone have any objections?