R Markdown parameterized pdf output using xelatex engine failed to compile when include ggplot

I'm using R Markdown to bulk generate parameterized pdf reports. I have no problem running each code chunk and no problem knitting single pdf. But when I run the lapply() function at the bottom, three (the "cut" variable in diamonds dataset has five categories) pdfs are generated successfully but I'm stuck at the fourth one, "Very Good". I get error messages instead:

WHAT DID I TRY:

I have re-installed R, RStudio, tinytex, and every package used in my code. I have pinpointed that the issue has something to do with the ggplot code chunk - hence, all five pdfs are generated successfully if I exclude that code chunk.

UPDATE:

1. As folks pointed out below, the breakage running lapply() function is caused by the blank space in "Very Good", and potentially, because of how Windows system cannot deal with blank space in path name. But these cannot explain why the code stuck at "Very Good" ONLY WHEN I INCLUDE A PLOT, in other words, when I take the plot code chunk out, the old lapply() code works for "Very Good" and other categories in my parameter "cut". Appreciate if somebody could get to the bottom of this issue.
2. A couple of us are getting a new error message regarding fontspec saying that fontspec requires "xelatex or laulatex" but I've already defined "latex_engine: xelatex" in my YAML. I'd really appreciate if folks could pinpoint this issue as well or let me know if there's any other way to change the default font to user defined font such as "Times New Roman" in another way to by pass this fontspec issue?

CODE:

The following is everything I have in my current "testing_for_stackoverflow.Rmd" file that results in the fontspec error message:

---
output: 
  pdf_document:
    latex_engine: xelatex
tables: true
graphics: yes
header-includes:
  - \usepackage{booktabs}
  - \usepackage{fontspec}
  - \usepackage{graphicx}
  - \usepackage{amsmath}
  - \setmainfont{Times New Roman} 
  - \newcommand{\bcenter}{\begin{center}} 
  - \newcommand{\ecenter}{\end{center}}
params:
  cut: "Very Good"
---
\fontsize{10}{12}
\selectfont

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE,
                      warning = FALSE,
                      message = FALSE,
                      fig.align = 'center',
                      include = TRUE,
                      tinytex.tlmgr_update = FALSE,
                      results='asis',
                      fig.dim = c(7, 2.5))
pacman::p_load(tidyverse)

```


## I'm using "diamonds" dataset and cut is my parameter

```{r}
filter(diamonds, cut == params$cut) %>%
  ggplot(aes(color, price)) +
  geom_boxplot()

```

```{r}
kableExtra::kable(head(filter(diamonds, cut == params$cut))) 
```

Run this code in the console to produce several output files:

library(tidyverse)
lapply(unique(diamonds$cut), function(i) {
  rmarkdown::render(input = "testing_for_stackoverflow.Rmd",
                    params = list(cut = i),
                    rmarkdown::pdf_document(),
                    output_dir = "out/",
                    output_file = paste0(i, ".pdf"))
})

The fourth document rendered (or rather not rendered) is the one containing a blank in the supposed intermediary and final filenames: Very Good.... Guessing from the error message, the filename might be falsely interpreted as two separate instructions (undefined control sequence).

Does replacing the blank solve the issue? One approach:

lapply(make.names(unique(diamonds$cut)), ...)

instead of:

lapply(unique(diamonds$cut), ...)

HOWEVER, make.names() will return "Very.Good" in the first argument of lapply() but since params = "Very Good", not data will be found in the rendering process. As a result, the pdf output of "Very Good" returns NA for all the data points being used.

So, the ultimate solution is to create a new variable using make.names(cut) and use the new variable as a parameter in the Rmd file for this parameterized report code to run (in a Windows system).