Looks like it is converted to <mi mathvariant="bold-italic-italic">&#x3B1;</mi>. At a fist glance it seems that there should not be italic in the mathvariant. According to the latest MathML standard mathvariant is only used to set and the corresponding unicode char 𝜶 should be used instead.

Change #1074320 had a related patch set uploaded (by Physikerwelt; author: Physikerwelt):

[mediawiki/extensions/Math@master] Fix heuristic that adds bold-italic

https://gerrit.wikimedia.org/r/1074320

It seems that it is sufficient to covert <mi mathvariant="bold-italic-italic">&#x3B1;</mi> to <mi mathvariant="bold-italic">&#x3B1;</mi>. Now the output looks fine with MahtJax

For sake of good form the test should really include other symbols,

\boldsymbol{\alpha \beta \gamma \delta \epsilon \zeta \eta \theta}
\boldsymbol{\iota \kappa \lambda \mu \nu \xi \omicron \pi}
\boldsymbol{\rho \sigma \tau \upsilon \phi \chi \psi \omega}
\boldsymbol{\varepsilon\digamma\varkappa\varpi}
\boldsymbol{\varrho\varsigma\vartheta\varphi}

all currently fail, some of the varients might cause problems.

Yes, it's all running the same code.

Change #1074320 merged by jenkins-bot:

[mediawiki/extensions/Math@master] Fix heuristic that adds bold-italic

https://gerrit.wikimedia.org/r/1074320

This should be fixed with the next deployment.