class CodecAdapterEvents

If the core codec needs to fail the whole CodecImpl, such as when/if new

FormatDetails are different than the initial FormatDetails and

the core codec doesn't support switching from the old to the new input

format details (for example due to needing different input buffer config).

The core codec should only call this method at times when there is a

current stream, not between streams.

The core codec should only call this method at times when there is a

current stream, not between streams. If the core codec calls this method,

the core codec must also override CoreCodecResetStreamAfterCurrentFrame(),

which _may_ be called async from the StreamControl thread, if the client

hasn't already moved onto a new stream by then.

This call requests a call to CoreCodecResetStreamAfterCurrentFrame() on the

StreamControl thread (async with respect to this method call), if the

current stream isn't obsoleted first.

Deprecated; use onCoreCodecMidStreamOutputConstraintsChange2 instead.

Passing false for output_re_config_required is deprecated (and has been for

a long time) and passing false will terminate the current process in

release builds as well as debug builds. The old meaning of false is now

communicated using onCoreCodecOutputFormatChange instead.

"Mid-stream" can mean at the start of a stream also - it's just required

that a stream be active currently. The core codec must ensure that this

call is properly ordered with respect to onCoreCodecOutputPacket() and

onCoreCodecOutputEndOfStream() calls.

When not using dynamic buffers, a call to

onCoreCodecMidStreamOutputConstraintsChange2 must not be followed by any

more output (including EndOfStream) until the associated output re-config

is completed by a call to CoreCodecMidStreamOutputBufferReConfigFinish().

When using dynamic buffers there is no such restriction - the core codec is

free to output a packet or EOS using any currently-added buffer (see also

EnableSameOutputBufferConcurrentlyInFlight).

The constraints_version value is in the same sequence as

CoreCodecGetBufferCollectionConstraints3Result.constraints_version.

When the core codec calls this method, the CodecImpl will note that the

format has changed, and on next onCoreCodecOutputPacket(), the CodecImpl

will ask the core codec for the format and generate and send an

OnOutputFormat() message before that output packet. This way, the core

codec is free to call onCoreCodecOutputFormat() repeatedly without any

packet in between, with CodecImpl collapsing these into one

OnOutputFormat() to avoid the extra message (so it doesn't have to be sent

and doesn't have to be handled by clients).

CodecAdapter(s) should take care to call onCoreCodecOutputPacket

referencing a buffer before any dropping of any CodecAdapter(s) handle(s)

to the referenced buffer.

This must only be called between onCorecodecInputPacketDone and onCoreCodecOutputEndOfStream.

This does not need to be called if CoreCodecStopStream is called before a timestamp_ish value

has been resolved as to whether it would have generated output.

If the CodecAdapter sub-class ever calls this method, the CodecAdapter sub-class must also

overide CoreCodecMetricsImplementation(), and not return std::nullopt from that method.