common-id extractor handles 'description' poorly

[Harald Sitter]

appdata.xml contains

```
<description>
<p>foobar</p>
</description>

```

Which is valid as per https://www.freedesktop.org/software/appstream/docs/chap-Metadata.html#sect-Metadata-GenericComponent

Snapcraft fails to extract this when using the adopt-info and parse-info feature. Supposedly because appstream.py contains

`description = _get_value_from_xml_element(tree, 'description')`

which would get the description node which renders to an empty string. What it probably needs to do is get the description element and then iterate all children of the description and convert them to the plaintext format expected by snaps.

This defect renders the description/summary extraction broken for just about all KDE appdata files as they all contain at least one paragraph. In fact, this likely affects every single appdata file in existence as paragraph wrapping is necessary for localization as explained in the aforementioned appstream documentation.

[Sergio Schvezov (sergiusens)]

sergiusens: git clone --branch Neon/unstable https://anongit.neon.kde.org/kde/kblocks.git
12:21 sergiusens: depending on what you want to do kbruch.git might be better though as it has no additional deps besides the build-snap. in either case you'll need `export SNAPCRAFT_PARTS_URI=https://metadata.neon.kde.org/snap/parts.yaml` for the remaining remote part
<sergiusens> Sergio Schvezov Thanks sitter
12:24 This for appstream desktop files
12:31 <sitter> Harald Sitter sergiusens: kbruch might be better then, it also has a <ul> in the description ;)

[Sergio Schvezov (sergiusens)]

https://github.com/snapcore/snapcraft/pull/2430