Documentation does not show iOS16 ops for conv and conv_transpose

## 🐞Describing the bug
I have an ML Program with a conv node. If the pad_type is set to `same` or `same_lower` I get an error saying the 'pad' input is required.

> coremlcompiler: error: Failed to parse the model specification. Error: Unable to parse ML Program: in operation node1_token_0: Required param 'pad' is missing

The [spec for conv](https://apple.github.io/coremltools/source/coremltools.converters.mil.mil.ops.defs.html#coremltools.converters.mil.mil.ops.defs.iOS15.conv.conv) in coremltools says 'pad' should only be specified if the pad_type is `custom`.

Is there some inconsistency between the spec and the parsing in coremlcompiler and 'pad' needs to be specified?

## System environment (please complete the following information):
- coremltools version:
- OS (e.g. MacOS version or Linux type): MacOS 13.6.4

## Additional context
Zipped model.mlprogram:
[model.zip](https://github.com/apple/coremltools/files/14109078/model.zip)

Start with the conv and conv_transpose specifications linked in the issue and inspect the attached model.zip to reproduce the missing-pad error for an ML Program using same or same_lower padding. Compare the documented iOS15 behavior with the requested iOS16 operations; done means the relevant documentation and behavior discrepancy are clearly resolved.

Written by the indexing model from the issue text.