Document the render options the gem actually reads

What does this MR do and why?

I compared the options table in the README with the RenderOptions struct in ext/gitlab_glfm_markdown/src/glfm.rs, and three things in the README do not match what the gem reads:

• escape_char_spans is not an option. The field is escaped_char_spans (glfm.rs:18), which is also how the only_escape_chars row spells it. to_html passes the options hash to the native render_to_html_rs, which reads it with serde_magnus::deserialize (ext/gitlab_glfm_markdown/src/lib.rs:10), and RenderOptions does not set #[serde(deny_unknown_fields)], so a key it does not know is ignored without an error: a caller who copies the name from the README gets no <span data-escaped-char> and no warning. The row has had that name since it was added in !37 (merged). I renamed it and moved it after escape, where it now sorts.
• alerts and subscript had no row. Both are fields of RenderOptions (glfm.rs:10 and glfm.rs:58) and both are passed to Comrak (glfm.rs:75 and glfm.rs:90). I added a row for each in its alphabetical place.
• The Usage section did not mention glfm: true. to_html reads that key itself (lib/gitlab-glfm-markdown.rb:38) and, when it is set, starts from GLFMMarkdown::GLFM_DEFAULT_OPTIONS and merges the options passed over it; without it, an option that is not passed is off, because RenderOptions derives Default and sets #[serde(default)] (glfm.rs:7 and glfm.rs:8). I added two sentences and a one-line example.

GitLab's own renderer (lib/banzai/filter/markdown_engines/glfm_markdown.rb in gitlab-org/gitlab) passes alerts: true and escaped_char_spans: true, so two of the three rows this changes describe options GitLab itself relies on.

After the change the table has one row for each of the 43 RenderOptions fields, the new rows keep the column padding, and the rows stay in the order the table already used.

This touches README.md only, so I did not add a changelog entry.

How to set up and validate locally

1. Install the published gem 0.0.45, the release tagged at the current main (c4575ff7):

   docker run --rm -it ruby:3.4 bash
   gem install gitlab-glfm-markdown -v 0.0.45
   irb

2. In irb, compare the old and the new option name, then the two options the README did not list, then glfm: true:

   require 'gitlab-glfm-markdown'
   
   GLFMMarkdown.to_html('a \* b', options: { sourcepos: false, escape_char_spans: true })
   # => "<p>a * b</p>\n"
   GLFMMarkdown.to_html('a \* b', options: { sourcepos: false, escaped_char_spans: true })
   # => "<p>a <span data-escaped-char>*</span> b</p>\n"
   
   GLFMMarkdown.to_html("> [!note]\n> Text\n", options: { sourcepos: false, alerts: true })
   GLFMMarkdown.to_html('H~2~O', options: { sourcepos: false, subscript: true })
   
   GLFMMarkdown.to_html("| a | b |\n|---|---|\n| 1 | 2 |\n", options: { glfm: true })
   GLFMMarkdown.to_html("| a | b |\n|---|---|\n| 1 | 2 |\n", options: { glfm: true, sourcepos: false })

3. Read the rendered README on this branch: the alerts, escaped_char_spans and subscript rows, and the paragraph under the first Ruby example.

Output from the published 0.0.45 gem (Docker ruby:3.4, Ruby 3.4.11, x86_64-linux)

1. escape_char_spans (README today) vs escaped_char_spans (what the gem reads)
1a: to_html("a \\* b", options: {sourcepos: false, escape_char_spans: true})
  => "<p>a * b</p>\n"
1b: to_html("a \\* b", options: {sourcepos: false, escaped_char_spans: true})
  => "<p>a <span data-escaped-char>*</span> b</p>\n"

2. alerts
2a: to_html("> [!note]\n> Text\n", options: {sourcepos: false})
  => "<blockquote>\n<p>[!note]\nText</p>\n</blockquote>\n"
2b: to_html("> [!note]\n> Text\n", options: {sourcepos: false, alerts: true})
  => "<div class=\"markdown-alert markdown-alert-note\">\n<p class=\"markdown-alert-title\">Note</p>\n<p>Text</p>\n</div>\n"

3. subscript
3a: to_html("H~2~O", options: {sourcepos: false})
  => "<p>H~2~O</p>\n"
3b: to_html("H~2~O", options: {sourcepos: false, subscript: true})
  => "<p>H<sub>2</sub>O</p>\n"

4. glfm
4a: to_html("| a | b |\n|---|---|\n| 1 | 2 |\n", options: {})
  => "<p>| a | b |\n|---|---|\n| 1 | 2 |</p>\n"
4b: to_html("| a | b |\n|---|---|\n| 1 | 2 |\n", options: {glfm: true})
  => "<table data-sourcepos=\"1:1-3:9\">\n<thead>\n<tr data-sourcepos=\"1:1-1:9\">\n<th data-sourcepos=\"1:2-1:4\">a</th>\n<th data-sourcepos=\"1:6-1:8\">b</th>\n</tr>\n</thead>\n<tbody>\n<tr data-sourcepos=\"3:1-3:9\">\n<td data-sourcepos=\"3:2-3:4\">1</td>\n<td data-sourcepos=\"3:6-3:8\">2</td>\n</tr>\n</tbody>\n</table>\n"
4c: to_html("| a | b |\n|---|---|\n| 1 | 2 |\n", options: {glfm: true, sourcepos: false})
  => "<table>\n<thead>\n<tr>\n<th>a</th>\n<th>b</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>1</td>\n<td>2</td>\n</tr>\n</tbody>\n</table>\n"

5. an unknown key is ignored without an error
5a: to_html("x", options: {sourcepos: false, no_such_option: true})
  => "<p>x</p>\n"

subscript beside strikethrough, which the new row describes:

"H~2~O ~~old~~" {sourcepos: false, strikethrough: true}
  => "<p>H<del>2</del>O <del>old</del></p>\n"
"H~2~O ~~old~~" {sourcepos: false, strikethrough: true, subscript: true}
  => "<p>H<sub>2</sub>O <del>old</del></p>\n"

Other checks

• The table's option names against the RenderOptions fields (the three commented-out fields left aside): on main, alerts and subscript are fields with no row and escape_char_spans is a row with no field; on this branch both lists hold the same 43 names.
• The rows still sort in byte order (LC_ALL=C sort -c over the option column), which is the order the table already used (escape before escaped_char_spans, tasklist_classes before tasklist).
• Rendering this branch's README with the gem itself gives 43 table rows, and the three new rows render with their code spans intact.
• The project runs no Markdown linter. With markdownlint-cli2 0.23.2 at its defaults, the only findings this branch adds over main are MD013 on the two added table rows, which every existing row also has.

Related to !37 (merged), !60 (merged) and !70 (merged), where escaped_char_spans, subscript and alerts were added.