feat(docs): Replace PDF download link with HTML zip download link

PDF build would randomly fail after S3 docs grew to a certain size.

At 3k+ pages PDF is not very usable anyways.
For offline use an HTML archive makes more sense.

docs/conf_common.py

@@ -317,6 +317,7 @@ extensions += ['sphinx_copybutton',
                'esp_docs.idf_extensions.kconfig_reference',
                'esp_docs.idf_extensions.gen_idf_tools_links',
                'esp_docs.esp_extensions.run_doxygen',
+               'esp_docs.esp_extensions.add_html_zip',
                ]
 
 # Use wavedrompy as backend, instead of wavedrom-cli

docs/en/conf.py

@@ -1,9 +1,9 @@
+# type: ignore
 # -*- coding: utf-8 -*-
 #
 # English Language RTD & Sphinx config file
 #
 # Uses ../conf_common.py for most non-language-specific settings.
-
 # Importing conf_common adds all the non-language-specific
 # parts to this conf module
 try:
@@ -25,3 +25,5 @@ copyright = u'2016 - {}, Espressif Systems (Shanghai) Co., Ltd'.format(current_y
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.
 language = 'en'
+
+html_zip = f'esp-idf-{language}-{release}'  # noqa: F405

docs/zh_CN/conf.py

@@ -1,9 +1,9 @@
+# type: ignore
 # -*- coding: utf-8 -*-
 #
 # English Language RTD & Sphinx config file
 #
 # Uses ../conf_common.py for most non-language-specific settings.
-
 # Importing conf_common adds all the non-language-specific
 # parts to this conf module
 try:
@@ -25,3 +25,5 @@ copyright = u'2016 - {} 乐鑫信息科技（上海）股份有限公司'.format
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.
 language = 'zh_CN'
+
+html_zip = f'esp-idf-{language}-{release}'  # noqa: F405