The reason why Adobe Acrobat Reader complains is because there indeed is trash in the content streams of the second and third page of your document, there are multiple sections like this
0 Tc
1 0 0 sc
? ? m
? ? ? ? ? ? c
? ? ? ? ? ? c
? ? ? ? ? ? c
? ? ? ? ? ? c
f
0 0 0 sc
BT
1 1 1 sc
/F1 10 Tf
? ? Td
(12) Tj
ET
All these question marks would have to be numbers to make this a valid PDF content stream. There even is one similar section with numbers filled in:
1 0 0 sc
18.5 443.384 m
18.5 447.25 21.63401 450.384 25.5 450.384 c
29.36599 450.384 32.5 447.25 32.5 443.384 c
32.5 439.51801 29.36599 436.384 25.5 436.384 c
21.63401 436.384 18.5 439.51801 18.5 443.384 c
f
0 0 0 sc
BT
1 1 1 sc
/F2 10 Tf
20 440.384 Td
(12) Tj
ET
I would assume that the producer here either used the '?' as placeholders in a template and simply forgot to fill-in these placeholders (probably what you have actually is a template, not a PDF for distribution or viewing) or added these sections with NaNs (not-a-number values) which then were serialized as question marks.