csvsimple inside tabular environment, but csv data includes LaTeX arrays/tables/aligned etc

I am trying to import a csv using csvsimple, and I have already overcome the challenge of the csv delimiter (by adding curly braces around every field), but I am now stuck with a admittedly weird problem. I would explain in more detail, but a simplified example shows this issue nicely enough.

I’ve adapted the example information from an answer found here, but separated the csv data to a standalone file.

let jobname.csv be:

11.0; 12.3; $x^2$; seven$
1.2; 33.11; five; six

let our .tex file be:

documentclass{article}
usepackage{booktabs}
usepackage{csvsimple}

begin{document}
begin{tabular}{cccc}
toprule
1 & 2 & 3 & 4 
midrule
csvreader[
  separator=semicolon,
  no head,
  late after line=,
]{jobname.csv}{}{%
  csvcoli & csvcolii & csvcoliii & csvcoliv
}
bottomrule
end{tabular}

bigskip

end{document}

When I run this everything is fine.

The issue happens when I include an array or some similar environment that uses & and .

Change contents of jobname.csv to:

11.0; 12.3; $x^2$; begin{array}{cc}2 & 4  6 & 8end{array}
1.2; 33.11; five; six

When I run this, everything is not fine. From what I can tell, the causes the problem, and then TeX tries to interpret the & as part of the tabular environment.

I understand this is similar to this question, but I fear automating the conversion of my csv data to fit this data would be quite difficult, since my actual project already contains some 300 cells of LaTeX snippets.

11.0; 12.3; $x^2$; $begin{array}{cc}2 & 4  6 & 8end{array}$
1.2; 33.11; five; six

documentclass{article}
usepackage{booktabs}
usepackage[l3]{csvsimple}

begin{document}

csvreader[
  tabular=cccc,
  table head=toprule 1 & 2 & 3 & 4  midrule,
  separator=semicolon,
  no head,
  late after line=,
  late after last line=bottomrule,
]{jobname.csv}{}{%
  csvcoli & csvcolii & csvcoliii & expandafterunexpandedexpandafter{csvcoliv}
}

end{document}