That "charter" is rather down-to-earth; it does not specify that "TLS should provide confidentiality and integrity" because this is taken to be obvious; instead, the charter is a roadmap to the future TLS 1.3 and thus documents the desirable changes from TLS 1.2.
As for the mailing-list messages you are pointing to, I think you are over-interpreting them. The kind of "middleware boxes" can be, for instance, a load-balancer system that needs to inspect
ClientHello messages so that it redirects session resumption attempts to the server which last handled that session. Or a RADIUS server using EAP-TLS as authentication protocol: in EAP-TLS, the EAP layer must recognize the record boundaries in order to wrap them into EAP messages. These are very legitimate, non-evil reasons for systems residing in-between the TLS client and server to inspect the packets and actually understand the unencrypted parts of the protocol.
The dialogue in the mailing-list goes mostly the following way:
We must encrypt all the things ! Our Charter demands it ! Let's change the record header format.
This will break systems which rely on the record format; this may break compatibility with middleware boxes (that I will allude to only in the vaguest terms). Is it really required ? Or is it a Gratuitous Change, that is explicitly forbidden by Our Charter ?
This is needed because content-type field in the header allows for evil traffic analysis attacks (that I won't describe because I don't actually know of such an attack). Traffic Analysis is Very Serious ! Encrypt all the things !
But Our Charter does not talk about Traffic Analysis ! And Traffic Analysis is much harder to defeat than simply encrypting the record content-type headers. Traffic Analysis is Out Of Scope !
and so on, ad nauseam. Interspersed in the exchange are actual bits of information, from which we can conclude that the WG members do not seem to have reached (yet) a consensus as to the level of backward compatibility that TLS 1.3 should have with previous protocol versions. Especially since that compatibility is defined relative to existing, deployed system that interact with the on-the-wire format in various way that are not necessarily well-documented.
However, reading these emails as if the WG was trying to promote some generalized NSA-like spying would be erroneous. This is a classic case of "committee reasoning" when the actual goals and constraints have been insufficiently specified. It can be predicted that the definition of TLS 1.3 will take some time (years) to come to fruition.