Indentation problem with wrapfigure and listings

Question

I am having a problem with some indentation of paragraphs when displaying an image next to a block of lstlisting. I am using Overleaf online (and have found several weird quirks previously). Below is a MWE.

\documentclass{article}
\usepackage[utf8]{inputenc}
\usepackage{wrapfig}
\usepackage{listings}
\usepackage{geometry}
\usepackage{setspace}
\usepackage{graphicx}
\usepackage{lipsum}
\geometry{letterpaper, portrait, margin=1in}

\lstset{
    basicstyle=\ttfamily,
    columns=fullflexible,
    keepspaces=true
}

\begin{document}

\doublespacing

\lipsum[1]

\begin{wrapfigure}{l}{0.5\textwidth}
    \centering
    \includegraphics[width=0.48\textwidth]{He3Defect.png}
    \caption{Amplitudes of the 3-component OP components}
\end{wrapfigure}

\begin{lstlisting}
NOTICE: using acceleration to solve.
Building problem...done
    time: 0 second(s).
solving...
    Solver: successfully built
    time: 0 seconds.
estimated: 21% done
Found solution:
    iterations = 42
    relative error = 7.61862e-07
    time: 11 second(s).
\end{lstlisting}
        
\subsection{Free Energy}
Found a converging value for the free energy in 1 dimension.

The free energy calculations for 2 dimensions have not yet been correctly coded.

\lipsum[1]

\newpage

\section{Analysis}
The program will often converge to the correct solution in a reasonably small amount of time.

The possible causes of error were observed to come from poor choices of the relaxation parameter, too small of an iteration limit, poor initial guess, or a bad combination of step size and domain.

The result of the 1-dimensional free energy integral suggests...

The code has not yet been written to model neither the stripe phase nor the polka-dot phase.

\lipsum[2]

\end{document}

And this is some of the resulting document:

It appears almost as if the image is very tall, except I have verified that it is not; the extra space can span onto a new page and it affects all pages after it

I read that \lipsum may cause some strange things as well (Problems with wrapfigure), but any text that I put in the same area does the same thing (so I just use \lipsum for brevity). I also tried several other suggestions (Problem with wrapfigure and line breaks in the text, wrong indentation after wrapfigure, How to wrap text around a figure?, and Textwidth wrong after wrapfigure), but no luck.

A page on Overleaf says,

Take care while using adding wrapfigures very near the top or bottom of a page, as this can often cause unwanted effects that are hard or near-impossible to solve. It is not advisable to try to use wrapfigures alongside equations or sectional headers. They also cannot be used in lists, such as itemize and enumerate environments.

Does lstlisting affect it in the same way?

Any ideas on how to fix it?

Answer 1

This shows how to use paracol for this. It is incredibly robust, but you have to manually split paragraphs. Just let paracol run too long and see where you want the split to occur.

The [h] option for figure is needed to add the gap at the top. I added the split a line early just to show how it is done.

\documentclass{article}
\usepackage[utf8]{inputenc}
\usepackage{paracol}
\usepackage{listings}
\usepackage{geometry}
\usepackage{setspace}
\usepackage{graphicx}
\usepackage{lipsum}
\geometry{letterpaper, portrait, margin=1in}

\lstset{
    basicstyle=\ttfamily,
    columns=fullflexible,
    keepspaces=true
}
% used to manually split paagraphs
\newsavebox{\textbox}
\newcommand{\splitpar}[2][\textwidth]{% #1 = width of column (optional), #2 = rest of paragraph after split
  \unskip\strut{\parfillskip=0pt\parskip=0pt\par}%
  \global\setbox\textbox=\vbox{\hsize=#1\relax\noindent\strut #2\strut}}
\newcommand{\continuepar}{\unvbox\textbox}

\begin{document}

\doublespacing

\lipsum[1]

\begin{paracol}{2}
\begin{figure}[ht]
    \centering
    \includegraphics[width=\linewidth, height=3.9in]{example-image}
    \caption{Amplitudes of the 3-component OP components}
\end{figure}
\switchcolumn
\begin{lstlisting}
NOTICE: using acceleration to solve.
Building problem...done
    time: 0 second(s).
solving...
    Solver: successfully built
    time: 0 seconds.
estimated: 21% done
Found solution:
    iterations = 42
    relative error = 7.61862e-07
    time: 11 second(s).
\end{lstlisting}
        
\subsection{Free Energy}
Found a converging value for the free energy in 1 dimension.

The free energy calculations for 2 dimensions have \splitpar{not yet been correctly coded.}
\end{paracol}
\continuepar

\lipsum[1]

\newpage

\section{Analysis}
The program will often converge to the correct solution in a reasonably small amount of time.

The possible causes of error were observed to come from poor choices of the relaxation parameter, too small of an iteration limit, poor initial guess, or a bad combination of step size and domain.

The result of the 1-dimensional free energy integral suggests...

The code has not yet been written to model neither the stripe phase nor the polka-dot phase.

\lipsum[2]

\end{document}