Make media pipeline storage more flexible

[Gallaecio]

Looking at how people need to use subclassing or monkeypatching for some straightforward cases of media pipeline storage, such as configuring the target Google Storage Cloud project ID or configuring a new storage class for a different service, I think we need to look into making media pipeline storage configuration more flexible.

Specifically, I think we need to consider:

• Expose FilesPipeline‘s STORE_SCHEMES as a setting (with a better name) that works also for images, rather than requiring subclassing to define new storage classes.
• Make storage classes Scrapy components, that can define from_crawler or from_settings to configured themselves based on settings, instead of requiring users to subclass base media pipeline classes to set class-level values based on settings, which feels like monkeypatching.