image/jpeg: misleading comment on maximum t value on receiveExtend() #81128

In current git repo (go1.27), the comment on huffman.go:receiveExtend() is the following:

// receiveExtend is the composition of RECEIVE and EXTEND, specified in section
// F.2.2.1.
//
// It returns the signed integer that's encoded in t bits, where t < 16. The
// possible return values are:
//
// - t == 0: 0
// - t == 1: -1, +1
// - t == 2: -3, -2, +2, +3
// - t == 3: -7, -6, -5, -4, +4, +5, +6, +7
// - ...
// - t == 15: -32767, -32766, ..., -16384, +16384, ..., +32766, +32767

But in scan.go:249, the DC path only rejects value > 16, so value == 16 is explicitly allowed.
That value is passed directly to receiveExtend in scan.go:251.

So the comment should indicate t <= 16 instead, and add

// - t == 16: -65535, -65534, ..., -32768, +32768, ..., +65534, +65535

(disclaimer: AI-found, Gemini and GPT seem to agree on t == 16 being allowed in the JPEG specs)