accept nodeBlock* delegates for ASCollectionNode

[fumito-ito]

for #22 and nits for ASBatchContext delegate

[dangthaison91]

LGTM. Thanks @fumito-ito

[vldalx]

@fumito-ito @dangthaison91
Had you tested this PR before you merged it?

According to ASTableDataSource docs tableNode:nodeBlockForRowAtIndexPath: method takes precedence over tableNode:nodeForRowAtIndexPath: if implemented.

According to the RP changes if configureCellBlock is not set then configureCell will be invoked from tableNode:nodeBlockForRowAtIndexPath: method. It means configureCell can be called on the main thread or a background queue. That is dead wrong.
Furthermore, two closures look redundant in that case.

[fumito-ito]

@vldalx thanks for your feedback.

It means configureCell can be called on the main thread or a background queue. That is dead wrong.

Is #14 same as your feedback ?

Furthermore, two closures look redundant in that case.

I cannot what you mean. Can you describe me more details for it ?

[vldalx]

@fumito-ito
take a look at the code below. it has the similar logic to the RP changes.

class ASTableSectionedDataSource {
    var configureCell: ConfigureCell
    var configureCellBlock: ConfigureCellBlock?

    init(configureCell: @escaping ConfigureCell) {
        self.configureCell = configureCell
        self.configureCellBlock = nil
    }

    func tableNode(_ tableNode: ASTableNode, nodeForRowAt indexPath: IndexPath) -> ASCellNode {
        // never invoked.
        // configureCell is supposed to called here
    }

    func tableNode(_ tableNode: ASTableNode, nodeBlockForRowAt indexPath: IndexPath) -> ASCellNodeBlock {
        let closure = configureCellBlock ?? configureCell
        // configureCell can't be fallback here
        return closure(self, tableNode, indexPath, self[indexPath])
    }
}

taking into account that approach, why do we need two closures? That just shows something is wrong.