Welcome to Audioholics Mike, especially to a DIY builder.
As mentioned above by everettT, a 3-way crossover is complicated. A 2-way is much more rewarding for a beginner. Your methods may have been sound on your previous 3-way center channel speaker, or you may have had beginner's luck.
To directly answer your question, wire two woofers (each one nominally 8 ohms) in parallel to achieve a nominal 4 ohms. Treat them equally while designing the crossovers. Don’t complicate things with a 3½-way or 4-way.
A hint: Variable L-pad attenuators may seem like a good idea, but they always seem to oxidize and fail soon. In the long run, it works better to do some trial and error with several different fixed value L-pad resistors until you like the balance across a crossover point. But this is among the last steps in making a good crossover.
How are you estimating the values of the crossover components? Are you using a crossover software application, or are you using text book type equations?
You do seem to understand that impedance in a speaker is not a constant value, it varies with the frequency. In the example below, a
SEAS Prestige ER18RNX, 4 traces are shown for that 6½" woofer. Three of them are frequency response curves measured on-axis (dark trace) and two angles off-axis (probably 15° and 30°). The driver’s impedance is graphed below them and uses the impedance vertical axis on the right side. It's the curve with the large peak between 30 and 40 Hz.
The manufacturers’ impedance curves are usually measured with the driver mounted on a large open baffle, not mounted on a cabinet. Madisound suggests some cabinet alignments:
Sealed box of 0.25 ft³ for an F3 of 97Hz.
Vented box of 0.5 ft³ with a 2" diameter vent × 6" long for an F3 of 53Hz. (I'd go with the vented box.)
If you want to use two of these woofers, build a test box where the woofer chamber is 1.0 ft³, where the front baffle is wide enough (>9") to mount two woofers one above the other, with two vents 2"×6". Mount the woofers and measure their impedance wired in parallel in that cabinet.
A properly designed crossover takes the variable impedance data into account rather than assumes one constant value. This is why I recommend using a crossover software application. This will also require a measuring rig.
You should do similar frequency response and impedance measurements for the mid range and tweeter as they are mounted on your test cabinet. The cabinet will have a noticeable effect on their responses as well.