The Remaining Length field is sometimes defined as a "IS a multibyte field" (see e.g. CONNECT packet line 623 or PUBLISH packet line 988) and sometimes as a "CAN BE a multibyte field (see e.g. PUBACK packet line 1171 or SUBSCRIBE packet line 1381).
I believe the wording should be more consistent and always define this field as a "Is a multibyte field".
I also noted that the Remaining Length field is not defined for the CONNACK packet, and incorrectly defined for the SUBSCRIBE packet (it should be "The length of the payload and the variable header.").