Add encode_lazy method to CodecContext

[rawler]

Some codecs (libvpx-vp9) can both buffer many frames, and take a long time encoding each frame. Accumulated, the last encode(None)-flush can end up taking a long time >30s, without detectable progress.

FFmpeg and many encoders themselves output one frame at a time, but PyAV currently buffer them all up into lists returned.

This change adds a encode_lazy(), yielding frames as they are made ready.

The change was benchmarked to also yield a net performance improvement. For both encode() and encode_lazy, encoding really small (24x18) frames using the mpeg4 encoder seem to take ~11% less time.

[rawler]

I also attempted to make the same change to decode(), but there I instead saw a clear (~10%) performance regression, so letting it be for now.

Benchmarked as

# ----------------------------------------------------------------
# Start encoding

for i in range(200):
    start = monotonic()
    with av.open("/tmp/test.mp4", mode="w") as container:
        stream = container.add_stream("mpeg4", rate=24, width=24, height=18, pix_fmt="yuv420p")
        for frame in frames:
            for packet in stream.encode(frame):
                container.mux(packet)

        # Flush stream
        for packet in stream.encode():
            container.mux(packet)

    encoding_times.push(monotonic()-start)

# ----------------------------------------------------------------
# Start encoding lazy

for i in range(200):
    start = monotonic()
    with av.open("/tmp/test.mp4", mode="w") as container:
        stream = container.add_stream("mpeg4", rate=24, width=24, height=18, pix_fmt="yuv420p")
        for frame in frames:
            for packet in stream.encode_lazy(frame):
                container.mux(packet)

        # Flush stream
        for packet in stream.encode_lazy():
            container.mux(packet)

    encoding_lazy_times.push(monotonic()-start)

Encode() elapsed before:

==== 200 samples ====
   0.107: ▌ (1)
   0.112: ███████████████████████▎ (43)
   0.117: ██████▌ (12)
   0.123: ███▊ (7)
   0.129: ███▊ (7)
   0.135: ████████████████████████████████████████████████████████████  (111)
   0.141: ██████████▎ (19)
  min: 0.105 avg: 0.129 max: 0.145
  perc: 95th: 0.141 99th: 0.144 99.9th: 0.144

Encode() elapsed after:

==== 200 samples ====
   0.107: ████████████████████████████████████████████  (66)
   0.112: ████████████████████████████████████████████████████████████  (90)
   0.117: ████████████  (18)
   0.123: █████▍ (8)
   0.129: █████▍ (8)
   0.135: ██████  (9)
   0.141: ▋ (1)
  min: 0.105 avg: 0.113 max: 0.145
  perc: 95th: 0.132 99th: 0.137 99.9th: 0.143

Encode_lazy() after:

==== 200 samples ====
   0.107: ████████████████  (31)
   0.112: ████████████████████████████████████████████████████████████  (116)
   0.117: ██████████▍ (20)
   0.123: █████▏ (10)
   0.129: █  (2)
   0.135: ████████▊ (17)
   0.141: ██▏ (4)
  min: 0.105 avg: 0.115 max: 0.145
  perc: 95th: 0.136 99th: 0.141 99.9th: 0.144

[WyattBlue]

Thanks, merging.