Do not escape dash, underscore, and ampersand

[kolyshkin]

Currently, we escape a few characters, such as

• - (dash) becomes \- (arithmetic minus sign)
• _ (underscore) becomes \_ (underline character)
• & (ampersand) becomes \& (zero-width space!)

As a result, we have two bugs:

1. All & characters disappear from the resulting man page.

2. Dashes are quite longer, and underscores becomes underlines (the
   latter are lower). This is not reflected in text output, only in PDF
   etc, and in most cases (like file names, command line options etc)
   does not look good.

In fact, the only character that needs escaping is the backslash itself.
This is what this patch does, together with fixes to expected test
output.

The following commands were used to generate the example PDFs:

$ bin/go-md2man < go-md2man.1.md > go-md2man.1
$ man -Tpdf ./go-md2man.1 > go-md2man.pdf

Before: go-md2man-old.pdf

After: go-md2man.pdf

Visual comparison:

Source for the above:

.PP
.RS

.nf
Old: go\-md2man \-\-example \-in=go\-md2man.1.md \-out=go\-md2man.1.out
New: go-md2man --example -in=go-md2man.1.md -out=go-md2man.1.out

Old: \-\-\- \_\_\_ \&\&\& k\_p\_q\_\_y\_z
New: --- ___ &&& k_p_q__y_z
.fi
.RE

[kolyshkin]

The "disappearing ampersand" issue can be seen in docker-run(1) man page.

Source:

   Start a container with a program to create a shared memory segment:
   ```
    $ docker run -it shm bash
    $ sudo shm/shm_server &
    $ sudo ipcs -m

man render:

       Start a container with a program to create a shared memory segment:

               $ docker run -it shm bash
               $ sudo shm/shm_server
               $ sudo ipcs -m

(or online at http://manpages.org/docker-run, search for sudo shm).

[kolyshkin]

@estesp @thaJeztah @cpuguy83 PTAL

[thaJeztah]

wondering if we actually need dependabot (there's only a single module we depend on, and pushing it to a newer version also forces consumers of this module to update)

[cpuguy83]

Its fine to have... except its already merged so looks like a bad rebase here.

[kolyshkin]

Not sure how it ended up here. Rebased, seems fixed now.

[thaJeztah]

sounds reasonable. I went looking in history to see if a specific reason was mentioned, but this function was part of the initial commit 6d7399d

[cpuguy83]

Yeah, I think the original implementation I was having problems with certain things and used that... indeed its been getting smaller and smaller.
Admittedly this was put together pretty quickly.

[cpuguy83]

LGTM

[estesp]

cool! I was looking at the && problem yesterday and was going to remove it from the escape list but you beat me to it :) And, did better by finding all the ones that didn't need escaping. Thanks @kolyshkin!

[thaJeztah]

Do you think we should do a new release?

[cpuguy83]

I was hoping to squash a couple more issues but the fixes weren’t quite right.
I went ahead and cut a 2.0.1 release.