Unlike Monier-Williams, who has plundered the wealth of Boehtlingk’s (& Roth’s) PW without proper acknowledgment of the materials on which he had based his structural rearrangement in the English language, the latter do acknowledge all of their sources and frequently quote them verbatim in extenso – whence it is called „Großes (big) PW“ [1853–75]: more quotes, less entries compared to „Kleines (little) pw“ [1879–89]: little or no verbatim quotes, but considerably more entries.

Using pw along with Richard Schmidt’s Supplement („Nachträge“) [1928] and Apte's Dictionary (which is truly independent) will yield a maximum of reliable results.

In the present case the sources of PW/pw were classical Sanskrit lexicographers, in one case the comparatively late Śabdakalpadruma.

> Are all or some of these 7 words in  Sabdakalpadruma and the Vacaspatyam?

The question can best be answered by consulting PW directly (sources highlighted in red colour):

kṣaudraja – (kṣaudra Honig + ja)  n.  Wachs  Râǵan. im Çkdr. (PW)

kṣaudrajā – f. Honigzucker Rājan. 14,133. (pw)

nihiṃsana – (von hiṁs mit ni)  n.  das Morden  Ak. 2, 8, 2, 81. (PW [+pw])

nītighoṣa – (nī° + ghoṣa)  m.  N. des Wagens des Bṛhaspati  Trik. 2, 8, 48. (PW [+pw])

nīcabhojya – (nīca + bho°)  m.  Zwiebel (die Nahrung gemeiner Leute)  Çabdaḱ. im Çkdr. (PW [+pw])

> rohaṇadruma were in MW but not PW or pw

This is wrong, for it is indeed recorded in PW/pw:

rohaṇadruma – m.  Sandelbaum  H. 641. Halâj. 2, 389. (PW [+pw])

> aṁsemūla

This again is inaccurate, first, because the correct entry in MW is aṃsamūla (not aṃsamūle), and second, because pw has registered it as a matter of fact (n. Schulterhöhe Rājan.)

I suggest to study first of all the introductions to the works of Boehtlingk & Roth to form oneself an idea of what exact methods in Sanskrit lexicography they had employed and who actually had contributed to their Herculean task.