Fix dimensionality demo for strict units

[ss2098]

This updates docs/examples/Dimensionality-Demo.py so that it runs with the current strict-units behavior.

The example previously failed because it:

1. created unit-bearing MeshVariable objects before reference quantities were set,
2. assigned plain dimensional values directly into .array, and
3. used older .to_nd(), .from_nd(), and UWQuantity reference-scaling calls that are not available in the current API.

The updated example keeps the dimensionality-tracking demonstration, but uses the current strict-units pattern:

• set reference quantities before creating unit-bearing variables,
• write non-dimensional values into .data,
• use scaling coefficients for dimensional/non-dimensional conversion.

No solver behavior is changed.

Tested with:

pixi run python docs/examples/Dimensionality-Demo.py

Result: script completes successfully.

[lmoresi]

Thanks @ss2098 — verified it runs end-to-end on current development (the dead to_nd/nd_array/from_nd API is correctly removed — those were dropped in the units simplification). Merging since a running demo is a clear win over the broken one.

Follow-up suggestion (non-blocking): since this is specifically the Dimensionality demo, it would showcase the units interface better to assign dimensional quantities via .array rather than manual value / scaling_coefficient → .data. I verified this works once reference quantities are set:

with uw.synchronised_array_update():
    T.array[...] = uw.quantity(1300.0, "kelvin")   # dimensional assignment
print(T.data[...])   # -> 1.3  (non-dimensional, == 1300/1000)

i.e. .array = dimensional (UnitAware), .data = non-dimensional — the manual scaling then isn't needed. Happy to push that polish as a follow-up if you'd like; not blocking this merge.

[ss2098]

Thanks Professor @lmoresi that makes sense. I created a small follow-up PR with the polished version here: #268.

It updates the demo to assign dimensional quantities through .array and then shows the corresponding non-dimensional values stored in .data, following your suggestion.

Kindly check when you have a chance and see if it works as intended.

Thanks,
Saurabh