Serde Deserialize+Serialize implementation

[rabbit-time]

This PR introduces a new feature flag: serde

I've implemented the Serialize and Deserialize traits using derive for ResolvedService and for all the local types which it contains.

Additionally, I've also written a basic test to accompany the changes and added serde_json as a development dependency.

If there is a benefit to providing serde support for types other than ResolvedService, please comment that.

Closes: #438

[rabbit-time]

CI jobs are currently failing since clippy isn't being run with features enabled.

Line 1286 in ./src/service_info.rs:
#[cfg_attr(feature = "serde", derive(Deserialize, Serialize))]

[keepsimple1]

Looks good! One inline comment related to the CI error. And optional, if you have bandwidth, you can also add a new step in CI to run cargo clippy --tests --all-features -- -D warnings to cover the features (I think).

[rabbit-time]

Also, one small thing to note. Currently, when there is no scope_id on an IPv6 address, it just puts in blank values.

@keepsimple1 My question to you is, should this be considered intended behavior as it reflects the internal representation with how the struct is implemented in Rust (uses default empty values as opposed to Option<T>) or should it be optimized for size and omit these fields in serialization?

[keepsimple1]

should this be considered intended behavior as it reflects the internal representation with how the struct is implemented in Rust (uses default empty values as opposed to Option) or should it be optimized for size and omit these fields in serialization?

I think we should keep it as is because scope_id is not Option in ScopedIpV6. It's empty here is because it's manually constructed in a test. It would not be empty in a real world service. (Or did I miss something?)

Btw, now ScopedIpV4 also got interface_id, which is the same as scope_id for IPv6. (see merged PR #439 ).

[rabbit-time]

I think we should keep it as is because scope_id is not Option in ScopedIpV6. It's empty here is because it's manually constructed in a test. It would not be empty in a real world service. (Or did I miss something?)

Btw, now ScopedIpV4 also got interface_id, which is the same as scope_id for IPv6. (see merged PR #439 ).

To my knowledge, an IPv6 address is not always accompanied by a scope ID. I may be wrong.

---

I will work on finalizing this PR today.

[rabbit-time]

Alright everything should be ready to go!

If you'd like, take one last look over everything. I moved the serde test into its own submodule so clippy would stop complaining about unused imports.

[keepsimple1]

LGTM! Thanks!