[types] Add completed field to ScanBlocksStart return type

[natobritto]

From v26 onward, the return type of scanblocks RPC has an argument called completed, this PR adds it and handles into type conversion and in previous versions changes.

Partially fixes issue #474.

[jamillambert]

It would be good to add an assert to the existing test that completed is returned for v26 and later. I checked the test for v26 and completed is true.

[natobritto]

Also included testing assert to ensure the completed field presence is verified from v26 onward.

[jamillambert]

ACK 39dfa65

[tcharding]

Perhaps this is better?

    let model: Result<mtype::ScanBlocksStart, ScanBlocksStartError> = json.into_model();
    let model = model.unwrap();

    let _: Option<ScanBlocksStatus> = node.client.scan_blocks_status().expect("scanblocks status");

    let _: ScanBlocksAbort = node.client.scan_blocks_abort().expect("scanblocks abort");

    #[cfg(not(feature = "v25_and_below"))]
    {
        assert!(model.completed.is_some());
    }

[natobritto]

Implemented, modified the model part for lint errors.

[tcharding]

Reviewed: 39dfa65

[tcharding]

Hit wrong button.

[tcharding]

We want to keep the error type here mate. The reason we do this which is admittedly strange is to make sure we exported the error type in the right place. We had bugs before we used this pattern when we forgot to do the re-exports.

[natobritto]

I see, introduced back in the type with slight name deviation for the lint.

[tcharding]

Reviewed 38e407d

[jamillambert]

Not a big fan of this _ just to get around the lint. Would it be better to assert something for all versions above this feature gated assert like assert!(model.from_height <= model.to_height)?

[tcharding]

I agree that we shouldn't use the under score, that has a specific meaning.