fix: Fixing images in the input Word files (#330)

* Fixing images identification in the input Word files

Signed-off-by: Maksym Lysak <mly@zurich.ibm.com>

* Populating extracted image data into docling picture for wordx backend

Signed-off-by: Maksym Lysak <mly@zurich.ibm.com>

* Updated tests

Signed-off-by: Maksym Lysak <mly@zurich.ibm.com>

* removed base64 dependency in msword_backend

Signed-off-by: Maksym Lysak <mly@zurich.ibm.com>

---------

Signed-off-by: Maksym Lysak <mly@zurich.ibm.com>
Co-authored-by: Maksym Lysak <mly@zurich.ibm.com>

tests/data/groundtruth/docling_v2/word_sample.docx.itxt

@@ -2,7 +2,7 @@ item-0 at level 0: unspecified: group _root_
   item-1 at level 1: paragraph: Summer activities
   item-2 at level 1: title: Swimming in the lake
     item-3 at level 2: paragraph: Duck
-    item-4 at level 2: paragraph: 
+    item-4 at level 2: picture
     item-5 at level 2: paragraph: Figure 1: This is a cute duckling
     item-6 at level 2: section_header: Let’s swim!
       item-7 at level 3: paragraph: To get started with swimming, fi ...  down in a water and try not to drown:

tests/data/groundtruth/docling_v2/word_sample.docx.md

@@ -4,6 +4,8 @@ Summer activities
 
 Duck
 
+<!-- image -->
+
 Figure 1: This is a cute duckling
 
 ## Let’s swim!