Give guideline of not using unit names in attributes (#1053)

Add a paragraph in the attribute name guidelines that discourage the use
of unit names and prefixes.

Add note about historical naming violation in messaging.md, the attribute names
message_payload_size_bytes and message_payload_compressed_size_bytes about the
incorrect use of bytes in the name.

Author: alexvanboxel

specification/common/attribute-and-label-naming.md

@@ -49,6 +49,10 @@ Names SHOULD follow these rules:
   name prohibits existence of an equally named namespace in the future, and vice
   versa: any existing namespace prohibits existence of an equally named
   attribute or label key in the future.
+
+- Names should not contain unit names; neither should it contain a fraction of 
+  the name's unit. For example, `http.request_content_length` does not have 
+  bytes in its name.
 
 ## Name Pluralization guidelines
 

specification/trace/semantic_conventions/messaging.md

@@ -181,6 +181,8 @@ For batch receiving and processing (see the [Batch receiving](#batch-receiving)
 Even though in that case one might think that the processing span's kind should be `INTERNAL`, that kind MUST NOT be used.
 Instead span kind should be set to either `CONSUMER` or `SERVER` according to the rules defined above.
 
+**Note:** The payload size names (`message_payload_compressed_size_bytes` and `message_payload_size_bytes`) don't comply with the [Attribute and Label Naming](../../common/attribute-and-label-naming.md), it should not contain `bytes`. This is due to historical reasons.
+
 ### Attributes specific to certain messaging systems
 
 #### RabbitMQ