Storage crate reorg and cleanup

[MindFlavor]

This is a seemly big PR but it's mostly code shuffling. With this PR I believe we are very close to release v0.1! 🥇

Purpose

1. Renamed the modules in blob_storage, table_storage, etc... to avoid module inception (see https://rust-lang.github.io/rust-clippy/master/index.html#module_inception).
2. Moved the blob_storage specific clients from core to the appropriate module.
3. Detangled the modules so the features compile independently (fixes Can't build azure_storage without default features or combinations of features less than the default #177). Right now by default it compiles everything but you can, say, use table_storage without anything else.
4. Removed hardcoded dependency from hyper (to support target wasm & wasi too #37). 🍾
5. Removed leftover dependencies no longer used.
6. Clippy'd everything. Most changes are really simple (ok_or_else etc...). The only "significant" change is the switch from Into to From. Now the crate is clippy-compliant. 🍾
7. Removed a lot of old code (most of it unnecessary, see below).

Cons

1. To remove the hyper dependency I had to remove the ability to generate signed URIs. We will reenable it in a future PR.

[rylev]

Looks good. I'm just unsure about the renaming of the submodules to have the _storage suffix. I'd like to hear the reasoning for that. I see you're trying to avoid module inception, but can you point to an example of that in the current code?

[rylev]

What was the reasoning behind adding the storage suffix to all the submodules?

[MindFlavor]

I've explained below, it's for the module inception: tight now there is a blob module inside the blob module. Now we have the blob module inside the blob_storage module (which I think makes sense)

[MindFlavor]

For example, the current master has a module table inside a module called table. Clippy complains with:

I cannot think of a better solution than appending storage but I'm open to suggestions.

[rylev]

@MindFlavor putting the the table model into a module called model would also fix this issue. I'm not against what you've done here, just trying to think about which solution is better.

[rylev]

😍 awesome!

[rylev]

Nit: I think the following is clearer:

s.strip_prefix('...').unwrap_or(s)