I feel like the direction here is an improvement, but even after reading through it a few times, it doesn't seem to quite be living up to a spell-fencer style, either. Maybe I'm just caught up on the idea of macro-rotation and wanting desperately to see multiple layers of synergetic ramp-up between spells and blade use. I feel like I'd usually be better able to articulate my critiques, but for now all I can say is that it still looks... incomplete?

I feel like the toggle on Spellblade will be very useful for this and similarly directed designs.

I'm not certain why you've left the mana gains on melee attacks so low. I'll have to compare these to the spell gains, but these still feel low. Like, swat-to-the-face low, where it'd sting less if they just gave you nothing. 2-4 seems like it'd be more reasonable.

*Nit-pick* I don't know if Flèche is the perfect word to tie a mana-crossing mechanic to, but it's at least a decent one. A more general Cross may be even better, in that the term can include the changing of sides about the opponent, or even just any of various terms in which one takes the inside or reorients the outside to become a new inside by side-stepping. But, I like the idea of the mechanic. I just feel like it, again, needs to control more mana, perhaps even the entirety of the last application's, to feel satisfying.

Manafication allows a shit-ton more opening burst than the mana-doubling. I can't honestly say which I'd prefer. So long as mana is being spent only on physical (albeit Enspell-ed) attacks, oddly enough, I think my preference lies with the doubling at a more frequent rate. Given a balance of throughput, I'd prefer to gain 50 mana each per 90s than 100 mana each per 3 minutes. Slightly. Very slightly.

I'll need to look at the throughput gains of the increased frequency of GCD or sub-GCD casts due to Clarity and especially Composure in order to give any numbers advice.

At this time we don't even know if auto-attacks will be intelligence-based—I'd assume not, given how many mechanics actively discourage them except during sword-spending—but should such sword integrations go through without increasing the AA speed, the difference between Composure and Clarity will be rather small. On a 6-second cast, it'd make .5 seconds lost uptime and 2.5 seconds lost auto-attack time (which may each be merely hitting for the equivalent of 2 potency). I'd include the shift to Intelligence-based AA damage in your notes (or have I just gone blind when reading wherever it's located?). That said, I'm not sure how I feel about auto-attacks consuming mana. In a pinch, it could just barely make you unable to use your last mana-consuming skill.