\boldmath selects the bold math version which would have bold versions of most or some of the math fonts set up in the "usual" math groups. However latin modern math has no bold font so this doesn't do anything useful here.
\boldmath just affects alphabetic characters and selects the bold roman text font.
Unfortunately unicode-math uses the command names for a rather different purpose and this is the main incompatibility between the current unicode-math definitions and classic LaTeX math definitions. Here
\mathbf is not changing the font but rather it is changing the characters within the same font so a is replaced by U+1D41A MATHEMATICAL BOLD SMALL A and similarly for the Greek (which would not be affected by
\mathbf in classic TeX)
This means that the bold effect is not achieved by resetting a font, or an assignment to a math
\fam but by stepping through token by token replacing tokens by alternative characters representing the bold. This means that while nothing's ever impossible hiding the explicit argument form here is likely to be tricky and fragile.
Other font sets such as stix may have more extensive bold support, allowing
\boldmath to work as intended, or using an explicit
\mathbf is probably the safest thing,