When using checkPythonSourceCode integrated in the source code editor,
for a scenario where developer edit a component and save we can benefit
from caching the check message for the source code content, because the
same check that the one happening in the editor will happen when the
component is saved.

This cache varies on:
 - "component_packages" cache cookie which is reset every time some
component code is edited.
 - zope startup time to take into account editions of file system code.
This assumes that after reseting file system code zope will be
restarted.
 - portal_type, because the checks performed by this function also
depend on portal type.

this makes completions works when using Ctrl+space

This makes "Format Document" / "Format Selection" work.

because pylint is a bit slow on large components, debounce every 2
seconds. TODO: this is too slow.

For now this is just the portal_url, but I'm thinking it could be a
proper tool.

IIRC the only thing needed is that ERP5TypeTestCase.getPortal is an
ERP5Site

on very large python files (>1000 lines) sometimes they queue up and we
have to wait for all requests that were queued by zope.

XXX maybe this does not happen when accessing through haproxy/apache, I
am observing this when hitting zope directly

See merge request nexedi/erp5!1512

See merge request nexedi/erp5!1522

Adding **kw prevents "Unexpected keyword argument 'select_favorite'".
The removed arguments url, user and password aren't used by the Script and can therefore be removed.

Allows using an empty fallback value with "replace" set to trigger header
removal.

See merge request nexedi/erp5!1501

erp5_openid_connect_client_login: Add dedicated cache factory to work with open id connect

erp5_openid_connect_client_login: Add Scope, Client Metadata and state

* Add configurable scope to connector
* Client Metadata is a JSON defined on connector
* Add state parameter to make redirect non repeatable

erp5_openid_connect_client_login: Update OpenId Connect to be be fully functionnal

erp5_openid_connect_client_login: Add tests and some minor fixup

erp5_openid_connect_client_login: Don't call create user on each call to callback

erp5_openid_connect_client_login: Test create connector in portal web services

erp5_openid_connect_client_login: Fix test on open Id connector creation in afterSetUp

OpenIDConnect: Add Extraction Plugin

openidconnect: Have functionnal Extractor

erp5_core: Add Case of OpenId Connect Client

erp5_xhtml_style: Add OpenId Connect to Login Form

erp5_web: Add OpenId Connect Logout

erp5_web_renderjs_ui: Add OpenId Connect to Login Form and Logout

erp5_credential: Add OpenId Connect to login form

It is GPDR compliant, as no cookie is set

Use CSP to improve protection against XSS and improve user privacy.

No javascript is supposed to be executed on the error page.

Such web sites are meant for anonymous access only, to improve CDN usage.

Do not load any JS, because the gadget may crash too, as no erp5 field is rendered

Allow users to go to the homepage.

Prevent img to increase the page width.

Most listbox functionnalities are not supported by web_js_style for now, as no form is used by default

Their contents is already provided by the HTML page template.

Generate a hfeed in the HTML rendering.
https://indieweb.org/feed

Provide a RSS feed automatically.

Only generate the feed when rendering the web section in view mode.

Extract all feed informations in Javascript to allow styling it as a blog.

Add bt5 dependency to erp5_dms to retrieve the author title.