A compelling answer is given in Rynasiewicz, R. (1996). Absolute Versus Relational Space-Time: An Outmoded Debate? The Journal of Philosophy, 93(6), 279-306. doi:10.2307/2941076
Isaac Newton provided the locus classicus for
substantivalism in the scholium to the opening definitions of the
Principia, where he laid out and defended the distinction between
absolute space and time and their relative counterparts. The major natural philosophers on the Continent, most notably Christian Huygens
and G. W. Leibniz, as well as such fringe figures for the new science
as Bishop Berkeley, voiced vehement objections, but failed to
offer any real alternative in the way of a dynamics founded on relationist
principles. Nor did any other classical relationist, including
Ernst Mach, succeed in this regard. (pp. 279-280)
Hence, although there were very good metaphysical reasons to refute absolute space, the mathematical tools of the time offered no means for a working alternative in relative frames so that the formulas could account for dynamic systems as Newton with his absolute space could. The same applies to Kant, who obviously was not the first (and not the last) one to refute this particular aspect of Newtonian physics.
In other words: Even though the philosophical insight was old in Kant's time already, Newtonian physics were the only ones offering formulas with predictive value for all aspects of mechanics known at the time. Obviously, even with a wrong premise, the conclusions still worked in non-relativistic environments (because of low velocity and/or low curvature of space), i.e. on earth and even within our solar system (mostly). It is only today that we are actually able to measure precisely enough to show relativistic effects in "common" cases.
Hence, Newton's mechanics were (and still are!) used in applications where the margin of error due to their not reflecting reality is so small that it does not matter for the practical considerations in question.