Hello Cydric,
Thank you for contacting the UCSC Genome Browser support team. I was able to replicate the same error that you are seeing. Since your file appears to match gff2 standards, I thought that there might be a problem with hidden characters.
I performed the following steps to fix the file:
First, I cut out all 9 columns:
cat UROD_Mb_geneid.gff | cut -f1,2,3,4,5,6,7,8,9 -d$'\t' > new.gff
I then replaced spaces with tabs:
awk -v OFS="\t" '$1=$1' new.gff
I was able to successfully upload "new.gff" as a custom track into hg38, as seen in this hg38 session.
I have also attached 'new.gff' with this response.