Add vscode.workspace.fs.createWriteStream().

[ericsnowcurrently]

(see #84175)

For the Python extension, we currently use node's fs.createWriteStream() for a variety of purposes:

1. downloading multi-MB files (and tracking progress)
   • calling into a third-party library that expects an fs.WriteStream
2. writing generated data out to a file a chunk at a time
3. writing log data out to a file, one entry at a time

For example:

• https://github.com/microsoft/vscode-python/blob/master/src/client/debugger/extension/adapter/logging.ts#L55
• https://github.com/microsoft/vscode-python/blob/master/src/client/common/net/fileDownloader.ts#L39

@jrieken

[ericsnowcurrently]

From #84175 (comment):

we have some proposed api but nothing is final yet. @ericsnowcurrently What is the used for? Real big files or just because?

See above.

[jrieken]

Let's use this issue to track the open, close, read, write API. That quite low-level but the building block for stream.

vscode/src/vs/vscode.proposed.d.ts

Lines 152 to 158 in ffe3749

	export interface FileSystemProvider {
	open?(resource: Uri, options: { create: boolean }): number | Thenable<number>;
	close?(fd: number): void | Thenable<void>;
	read?(fd: number, pos: number, data: Uint8Array, offset: number, length: number): number | Thenable<number>;
	write?(fd: number, pos: number, data: Uint8Array, offset: number, length: number): number | Thenable<number>;
	}

[bpasero]

From my recent experiences from implementing faster read/write IO for remote connections, I believe that the primitives are not a good choice and would rather swap them by a stream solution that has similar support (e.g. start offset + length).