rework list of predefined encodings

[milyin]

Describe the release item

There are several predefined encodings which became ambigious after serialization rework in #1474. The main reason of ambiguity is that the prefix ZENOH_ for encoding doesn't say anything is serialization used or not.

E.g. is ZENOH_STRING means that ZBytes::to_string() was used or string was serialized into ZBytes. The difference is important: in second case the string is additionally prefixed by length.
Same for ZENOH_BYTES value: is it ZBytes::to_bytes() or serialization of vector of u8. Difference is the same: just to_bytes() doesn't add any prefix

The proposal is to

• keep encodings ZENOH_STRING and ZENOH_BYTES, they should correspond to ZBytes::to_string(), ZBytes::to_bytes()
• add type ZENOH_SERIALIZED. it's suffix can be used to store serializing schema
• remove types ZENOH_INT8, ZENOH_INT16, etc

The fixes should be also propagated to bindings

• Encoding and timestamp serialization remove from zenoh-ext #1520
• zenoh-c
• zenoh-pico
• zenoh-cpp
• zenoh-python
• kotlin
• typescript

[milyin]

@wyfo @kydos if this that you finally decided about this issue?