Are you sure your genomes are exactly the same length without any insertion/deletion messing up the coordinates?

If yes some Perl code like this will work:

#!/usr/bin/perl

use strict;
use warnings;

my $start = 103538;
my $end = 105080;

$/="\n>"; # read Fasta sequences in blocks
while (<>) {
    s/>//g;
    my ($seq_id, @seq) = split(/\n/, $_);
    my $seq = join "", @seq;
    my $subseq = substr($seq, $start - 1, $end - $start);
    print ">$seq_id\n$subseq\n";
}

Usage:

perl getRegion.pl < INPUT_FASTA > OUTPUT_FASTA

This is trivial to do with Genozip:

genozip test.fa 
genocat test.fa.genozip -r JQ673480.1 --no-header --sequential | cut -c 103538-105080

Documentation: https://genozip.com

Paper: https://www.researchgate.net/publication/349347156_Genozip_-_A_Universal_Extensible_Genomic_Data_Compressor

see if this works with awk (assuming that fasta sequence records are in single line and each fasta record has the the required sequence within it):

$ awk -v OFS="\n" '/>/ {getline seq} {print $0, substr(seq,103538,1543)}' *.fa > new.fasta

or

$ awk '/>/; !/>/ {print substr(seq,103538,1543)}' *.fa > new.fasta

If fasta sequences are not in single line, try seqkit:

$ seqkit --quiet -w 0 subseq -r 103538:105080 *.fa > new.fasta

if you want to append gene name at the end of all fasta headers, do this:

$ seqkit --quiet -w 0 subseq -r 103538:105080 *.fa   | seqkit replace  -p "$" -r "_JQ673480.1" > new.fasta