HadoopStreams to support ByteBufferPositionedReadable input streams

### Describe the enhancement requested

If a stream declares in its StreamCapabilities that it supports
ByteBufferPositionedReadable, then use it for `readFully(ByteBuffer)`
All streams in Hadoop 3.0.0 + do declare this.

+ use StreamCapabilities to look for `ByteBufferReadable`.

For detecting ByteBufferReadable, use this probe falling back to the recursive scan.
All streams in the hadoop codebase will report this via StreamCapabilities, but there
may be some third-party streams which do not.

### Component(s)

_No response_

Start at the HadoopStreams implementation and trace readFully(ByteBuffer), then inspect how StreamCapabilities and the existing recursive scan are used to detect readable streams. Done means streams advertising ByteBufferPositionedReadable use that capability for readFully(ByteBuffer), while third-party streams still fall back to the recursive scan.

Written by the indexing model from the issue text.