Avoid the boilerplate in enums with strum

[Apostoln]

The goal of this PR is refactoring to avoid boilerplate in enums implementation.

First commit: Use strum crate for enums

Changes:

• Replace explicit impl Display with strum::Display
• Replace explicit impl FromStr with strum::EnumString
• Replace explicit impl From<Enum> for &'static str with strum::IntoStaticStr
• Add new error variant for strum::ParseError
• Replace explicit impl Default with derive

Changes in public interface:

• Now the error type of FromStr is strum::ParseError instead of BlockUtilsError(String). The reason is current limitation in strum, see https://github.com/Peternator7/strum/issues/91
• The Display string for DeviceState::FailFast is "failfast" instead of "fail_fast", the same as its value for parsing in FromStr. If it's not acceptable, please let me know and I will come back the explicit implementation.

Note:

• Unfortunately, I've overestimated the power of the strum crate. It has a lot of limitations (at least for now), therefore I've kept explicit impls in some cases.
• Strum can't handle correctly a case with default variant in Display and ToStr, like FilesystemType::Unrecognised(String). See https://github.com/Peternator7/strum/issues/86
• It seems I can't specify different attributes for serialization and deserializaition like "failfast" and "fail_fast" for DeviceState::FailFast, despite I don't see the issue for it.

Second commit: Use thiserror crate for errors Changes:

• Use derives of thiserror to avoid explicit impl Error, impl Display, impl From<OtherError> and so on.
• This change also allows to use strum for a bit more enums

This commit is not in the scope of initial issue, so I will revert it if you want.

Checks:

1. Current tests are Ok (but seems they don't use the changed functionality)
2. I've manually checked the accuracy of changes in terms of the same strings before and after these changes.
3. I use the fork with these changes in our project, and its tests are passed If you think that this PR can't be merged without adding new tests let me know.

Also it would be great to release a new version of block-utils after merging this PR due to changes in #25

Close #26