The Textual Witnesses
"Joseph" is found in some translations because the Greek word for Joseph (Ἰωσὴφ), is found in the majority of extant manuscripts. The majority text reading is reflected in the "RP Byzantine Majority Text," the two "Textus Receptus," and the "Greek Orthodox" renderings shown in parallel at Biblehub. Verse 33 is this ("Joseph" bolded):
Καὶ ἦν Ἰωσὴφ καὶ ἡ μήτηρ αὐτοῦ θαυμάζοντες ἐπὶ τοῖς λαλουμένοις περὶ
αὐτοῦ.
This is the text earlier English translations followed. The NA28 apparatus lists these witnesses as:
A K N Θ Ψ ƒ13 33. 565. 579. 892. 1424. 2542. l 844 𝔪 it vgmss syp.h
bopt
The 𝔪
in that list represents the hundreds or thousands of Byzantine manuscripts that make up the majority witness to this reading in Luke (I am not certain the number of manuscripts that contain Luke 2:33 exactly).
Notice in other renderings on that page that the text at the same location reads as the following (and one includes the additional bracketed word as well; the replacement words "the father of him" are bolded):
καὶ ἦν ὁ πατὴρ αὐτοῦ καὶ ἡ μήτηρ [αὐτοῦ] θαυμάζοντες ἐπὶ τοῖς
λαλουμένοις περὶ αὐτοῦ.
More recent English translations follow the minority witness reading because it is found in older extant texts. The NA28 apparatus lists these witnesses as:
א B D L W 1. 700. 1241. l 2211 vg sys.hmg sa bopt; Orlat
Why "Joseph"?
The earlier Greek text compilers and the early English translators were aware of many (though not all) of the variants found in extant texts today. Given that the Latin Vulgate (vg
above) has manuscripts with both readings, Erasmus would have likely been aware of the variant, and probably even in some of the extant Greek texts he had viewed in his research.
While there is some dispute about the number of texts he used for his Greek, they essentially represented the textform from the majority group. Why this was the case may also be disputed, but I would advance that it is because he valued the majority reading from the witnesses he was aware of, as some still do today, whereas many modern scholars give preference to the earlier dated extant witnesses (which are not necessarily the earlier textform, just representative of there being a textform of that type at an earlier time).
So those early English translations that you listed primarily used the early Textus Receptus versions based largely on Erasmus's work; but even in the case of the committee members themselves for the KJV translation of the Gospels are likely to have been aware of the variants, and chose to go with the majority reading (though a variant marginal note in chapter 2 would have entirely confirmed that, but they only selectively put variant notes if they felt a reading might be valid).
In short, the philosophy of determining which textform is correct has changed over time, and many later English translations follow the newer philosophy of determining the Greek text behind them that emphasizes extant older readings, whereas older English translations (and some modern ones) follow a different philosophy that emphasizes majority readings.