Step 1: Set up the two distinguishing parameters.
Any imaging sensor can be classified along two axes: the number of spectral bands $n$ it records, and whether those bands are contiguous (touching, so together they sample the spectrum almost continuously) or non-contiguous (separated by unused spectral gaps), together with the typical bandwidth $\Delta\lambda$ of each band.
Step 2: Place multispectral, hyperspectral and panchromatic sensors on this scale.
Panchromatic sits at $n = 1$, one very wide band. Multispectral sits at roughly $n \approx 3$ to $20$, non-contiguous bands, with $\Delta\lambda$ typically tens of nanometers to a few tenths of a micrometer. Hyperspectral sits at $n$ in the hundreds, contiguous bands, with $\Delta\lambda$ of only a few nanometers to about ten nanometers, giving near continuous spectral coverage.
Step 3: Classify each option using this scale.
(A): $n = 10$, non-contiguous, $\Delta\lambda = 0.04\ \mu m = 40\ nm$, this falls squarely in the multispectral range on all three counts. (B): $n = 256$, contiguous, $\Delta\lambda = 5\ nm$, both the large contiguous band count and the very narrow bandwidth place this in the hyperspectral range. (C): $n = 1$, $\Delta\lambda = 0.4\ \mu m$, a single band means this is panchromatic, since a sensor needs at least a handful of bands to be called multispectral. (D): $n = 1000$, contiguous, $\Delta\lambda = 1\ nm$, an extremely fine contiguous band set that is deep in hyperspectral (near ultraspectral) territory.
Step 4: Conclusion.
Testing $n$, contiguity, and $\Delta\lambda$ together, only option (A) satisfies all three multispectral criteria simultaneously, so it alone is correct.
\[ \boxed{\text{Option (A) only}} \]