Could you build a quasi-PWM MEME motif table file from your telomeric repeat motif, and then use FIMO to scan your contig FASTA for hits of various quality (q-value/p-value)? FIMO is kind enough to look for reverse-strand hits, as well. For instance, here's an example of how to search human assembly for TF motifs, which could be extrapolated to your situation with some alterations to inputs: bioinformatics.stackexchange.com/a/2491/776
The sequence is short enough, and your contigs likely short enough, that another option might be to programmatically build a set of off-target sequences (one mismatch, two mismatches, etc., along with their reverse complements) and search for those as regex patterns via a command-line script (using
re library in Python).
If you get same-strand hits back that are spaced every N bases (N being the length of the motif, e.g. N=6) then you may have located a telomeric repeat.