Optimize string concatenation using << operator

This change enhances the efficiency of HTML string construction, by
replacing `+=` operator with `<<`.

The shovel operator (`<<`) modifies strings in-place, reducing memory
allocation and improving overall performance compared to the `+=`
operator.

### Test Case

Ruby 3.3.5 x64 on M1 Pro

```rb
Arbre::Context.new { div { |d| d.ul { li } } }.to_s
```

#### Before

```
--- Memory Profiler ---

Total allocated: 4624 bytes (80 objects)
Total retained:  0 bytes (0 objects)

allocated memory by gem
-----------------------------------
      4624  arbre/lib

allocated memory by file
-----------------------------------
      2560  arbre/html/tag.rb
      1120  arbre/element/builder_methods.rb
       400  arbre/element_collection.rb
       280  arbre/element.rb
       264  arbre/context.rb

--- Benchmark: IPS ---

                test     50.546k (± 1.4%) i/s -    254.550k in   5.037005s

--- Benchmark: Memory ---

                test     6.208k memsize (     0.000  retained)
                        99.000  objects (     0.000  retained)
                        29.000  strings (     0.000  retained)
```

#### After

```
--- Memory Profiler ---

Total allocated: 3840 bytes (64 objects)
Total retained:  0 bytes (0 objects)

allocated memory by gem
-----------------------------------
      3840  arbre/lib

allocated memory by file
-----------------------------------
      1776  arbre/html/tag.rb
      1120  arbre/element/builder_methods.rb
       400  arbre/element_collection.rb
       280  arbre/element.rb
       264  arbre/context.rb

--- Benchmark: IPS ---

                test     52.205k (± 1.6%) i/s -    265.659k in   5.090028s

--- Benchmark: Memory ---

                test     5.424k memsize (     0.000  retained)
                        83.000  objects (     0.000  retained)
                        19.000  strings (     0.000  retained)
```

Author: tagliala

lib/arbre/html/tag.rb

@@ -116,23 +116,23 @@ def closing_tag
       def indent(open_tag, child_content, close_tag)
         spaces = ' ' * indent_level * INDENT_SIZE
 
-        html = ""
+        html = +""
 
         if no_child? || child_is_text?
           if self_closing_tag?
-            html += spaces + open_tag.sub( />$/, '/>' )
+            html << spaces << open_tag.sub( />$/, '/>' )
           else
             # one line
-            html += spaces + open_tag + child_content + close_tag
+            html << spaces << open_tag << child_content << close_tag
           end
         else
           # multiple lines
-          html += spaces + open_tag + "\n"
-          html += child_content # the child takes care of its own spaces
-          html += spaces + close_tag
+          html << spaces << open_tag << "\n"
+          html << child_content # the child takes care of its own spaces
+          html << spaces << close_tag
         end
 
-        html += "\n"
+        html << "\n"
 
         html
       end