The original question was 'inspiration'. The URLs provided are all hacker-written clients with a lite focus in active or recent development. I think they're a good place to look for inspiration because of the relative heterogeneity they represent across time, space, use case and architectural conception (some are layered or adjunct UXs, not monolithic integrated clients per-se). It is true that you could just as easily include WIMP GUI, private-hostable or cloud web UIs in there, but I would not anticipate such heterogeneity in those client categories save gimmicky minor feature improvements brought about by or serving edge cases like translation, mail(spam) merge, weird corporate office uses for email (like printing and file transfer), etc. You could also broaden the scope to general UX principals and metrics, ML applied to UX, mail-related service providers and tangential services such as hosting, address validation, identity verification, on-disk formats, non-SMTP store-and-forward messaging systems and their general evolution, non-store-and-forward messaging systems, and either oddball (printer touchscreen), arguably dominant (mobile phone) or upcoming UX formats (watches, AR/VR, OSX-style OS-managed merged notification stream systems). But the question implied a traditional computer and email UI.

I personally dislike the basic assumptions made by email clients such as assuming we want to organize inbox by date. This leads to vicious responses like inbox zero because the psychological effect and UX utility of an ever-growing list is basically negative. When considered temporally it is actually amazing how long it took major email UX to begin to fix this, and it's still basically broken. That's largely an artifact of the fact the specification is SMTP (T for Transfer) and thus the UX for end-user use case was never considered during protocol design. Today we would seek a more holistic design process. IMHO a strong novel UX would include integration support for digital forms of communication segregated from email such as file transfer, video conferencing, hosted issue systems, payment, etc. while recognizing the critical challenges of such an approach from a security perspective. You ideally need a walled garden, in other words, with OS level controls. I am skeptical a pure browser experience can deliver this. You really need an OSX-like Unix with user focus tracking, security guarantees and a managed notification stream or better.