Normalize SVD names to upper case where they were to pascal case

[whitequark]

Fixes #76.

Depends on calebmer/inflections#2.

[whitequark]

So... yes, I understand that right now, I am being that person who asks to change style conventions. However, I feel that there are a few factors in the favor of this change:

• The current behavior is clearly broken, and cannot be fixed and still be fully compliant with Rust's naming conventions for struct names (as described in Name mangling is lossy around numbers #76);
• Normalizing peripheral and register names to PascalCase is unprecedented in embedded development;
• Normalizing peripheral and register names to PascalCase means there is no trivial mapping from the datasheet names to API names (unlike with snake_case, where it is a simple substitution);
• Virtually all peripheral and register, and most enumeration names are complex abbreviations, which means that normalizing them to PascalCase always decreases readability;
• svd2rust already has precedent for this: 7b585ac.

And I suppose if this PR is merged, then the corresponding change should be done to cortex-m as well.

[japaric]

Normalizing peripheral and register names to PascalCase is unprecedented in embedded development;

(Rust is unprecedent in embedded development :-). And C just uses integers to represent registers instead of newtypes so C code doesn't have to deal with this.)

I kinda agree with the rationale outlined here but I'd like to take my time in assessing the breakage this will cause to dependent crates.

[whitequark]

Rust is unprecedent in embedded development :-)

I think you'll find the Habit bitdata types (see Habit report p.40) quite familiar, actually...

[whitequark]

@japaric From your latest post:

When using the RTFM framework you have to declare all the peripherals you are going to use in this macro. Both the name of the peripheral (for example: GPIOE) and its type (for example: Gpioe) must match the device crate, stm32f30x in this case, definitions.

Seems like this PR would make that a bit cleaner :)

[japaric]

Bascially r+ but we have to wait for inflections PR to be merged. Turns out this change won't break cortex-m-rtfm because the register_block field in the peripherals! macro can be made optional. (see japaric/cortex-m-rtfm#28)

[japaric]

This is waiting on a new release of the inflections crate.

[japaric]

Thank you @whitequark!

@homunkulus r+

[homunkulus]

📌 Commit 601cc70 has been approved by japaric

[homunkulus]

⌛ Testing commit 601cc70 with merge ae89a86...

[homunkulus]

☀️ Test successful - status-appveyor, status-travis
Approved by: japaric
Pushing ae89a86 to master...