Hi @mshelomi,
Sorry you're still stuck on this and thanks for providing the manifest file!
There are 2 things I can see that would get in the way of a successful import at the moment.
If this is correct, then your files are not gzipped, however the file-paths you provided in your manifest are looking for gzipped format files. For this you need to either gzip your individual files first, or correct the manifest file by removing the .gz extensions. The manifest type import will then automatically gzip the files itself.
The second thing I noticed was that in your manifest file your sample-ids are all unique instead of being paired. For example
A02T0-1 <- forward reads
A02T0-2 <-reverse reads
need to have an identical sample-id (ex. A02TO) so the script knows they belong to the same sample.
When these are fixed the following command should work 
:
qiime tools import \
--type 'SampleData[PairedEndSequencesWithQuality]' \
--input-path MANIFEST.csv \
--output-path pe-demux.qza\
--source-format PairedEndFastqManifestPhred33
Here I am guessing that your format is Phred33 as most recent illumina sequencers are, but you could double check that with your sequencing facility/operator.
The metadata file which provides environmental information about your samples is not required at the import step. Instead it is provided separately downstream as needed, depending on the analysis you are looking to do. More details, including its formatting and validation are explained here.
Hope this helps!