Do you remember when cryptography export was controlled? It was implemented by limiting key size to certain number of (effective) bits (of security). This suite is just a victim of that law, as it is a 1990s design.
It's not "just" a victim of that law unless they disclosed that the export cryptography protocol was trivially breakable. Export cryptography in the 1990s US was documented.
> The algorithm was developed in 1996/97 at Philips Crypto BV in Eindhoven (Netherlands) as a consultancy job for ETSI-SAGE. As the algorithm is secret, it has never been submitted for peer-review or in-depth security analysis. Instead it was evaluated by other ETSI-SAGE members before being submitted as a formal ETSI standard. All members of the TEA family, use an 80-bit key, but in the case of TEA1 it is effectively reduced to 32 bits, which makes it vulnerable to a brute-force attack. According to one of the developers, this was mandatory to get the algorithm approved for export. It was part of the ETSI specification and was clearly visible in the code [3].
American encryption laws did not exist in a void. According to this website, one person who was working on the standard indicated that the key space had to be reduced to allow for export.
French cryptography exports required authorisation if the key strength was higher than 40 bits. With its 80 bit keys, the TETRA key space would've been too big to qualify for free exports.
As TETRA is part of an ETSI standard, it seems pretty likely to me that one of the European countries had a 32 bit restriction, and TETRA might as well pick the lowest common denominator when it comes to selecting a backdoor.