Introduce a replaceable RestClient

[dangeross]

This PR:

• Introduces a RestClient trait (with ReqwestRestClient and MockRestClient implementations)
• Allows external REST services (e.g. mempool.space, LNURL services) to be mocked for testing
• Removes the mockito dependency, simplifies WASM testing and removes URL injecting for test purposes

[danielgranhao]

Awesome! I gave it a first pass and looks good. Will give it a second look later.

[JssDWt]

Looking good!
I like the way the responses can be set in the mock client. I very much like that we're not calling mempool.space in the tests anymore.

Some nits.

[JssDWt]

The docs for the reqwest client say

You do not have to wrap the Client in an [Rc] or Arc to reuse it, because it already uses an Arc internally.

So nit: I think functions can use a borrowed rest client rather than an Arc. That saves clones on passing it to the next function.

[dangeross]

Is this what you had in mind? 0f4633d

[JssDWt]

Not really. I had in mind

pub async fn validate_lnurl_withdraw(rest_client: &dyn RestClient)

... 

let mock_rest_client = MockRestClient::new();
validate_lnurl_withdraw(&mock_rest_client, withdraw_req, invoice);

But coding it up it would require generics, to support cloning the rest client. It would get quite some feet in the earth. So let's stick with the arcs.

[JssDWt]

What caused you to add this here in this PR?

[dangeross]

The MockRestClient was picking up a request to get LNURL info for 03864ef025fde8fb587d989186ce6a4a186895ee44a926bfc370e2c366597a3zzz@sdfsffs (in the node_id parsing trst). I think TLDs need a dot right?

[JssDWt]

Technically it could be localhost, or other domains in the local network. So this failed before because the network call failed?

[dangeross]

So this failed before because the network call failed?

Yes. Ok, it maybe makes sense to remove this then and add a mock response for this case