The vote of FESCo is highly critical: inappropriate outcomes can cause immediate harm to the community and Fedora Linux users, without the possibility to mitigate → it can render Fedora Linux unreliable and unpredictable. This is not the case with the Council, which shall add sufficient voice to the major community: it has a major impact and is the highest entity, everyone can vote, but they have not the potential to break Fedora Linux in the complex development and engineering processes throughout the many interrelations with other communities and projects.
This is not just to keep trolls out, but people who ain’t experienced in very technical open source decision making: in the critical/complex areas, open source depends on much discipline and understanding of the own limitations → leading to a dominant dynamic between rough consensus decision making (as people tend to focus on what they have knowledge about in order to not risk their reputation), reputation (to convince towards rough consensus) and the need to convince (for achieving rough consensus & adoption of the preference by many developers/engineers who need to adopt in order to avoid splits).
E.g., achieving TL3 is comparably easy, and it does not prove to understand highly technical decisions, nor qualification to understand who is best suitable for current technical challenges and who is embedded with whom (upstream) and who has good standing among what developers in order to achieve adoption of decisions (keep in mind that FESCo deeply depends on reputation rather than having a power to enforce). Open Source Developing is complex in areas that need review and that have many dependencies.
Also, keep in mind that TL3 cannot be held responsible, as they are not known: FESCo is the major entity that can/shall ensure to identify / mitigate “unqualified or malicious accounts” that can/want cause harm, e.g., when doing proposals. Giving FESCo voting power to TL3 would enable an easy attack on Fedora → if everyone can contribute to FESCo ballots just by having an FAS account, the outcome can be unpredictable, and Fedora development could end up with what we used to call “netsplit” in IRC (especially as FESCo can end up composed of people who do not have the qualification and/or standing to keep development and engineering together) 
Overlapping: this community is big and goes far beyond Discourse, and it is interwoven with many other communities and projects, on whose collaboration it depends. TL3 proves understanding of only a very small (and potentially isolated) share of that.
Additional entry barriers of FAS ensure that this cannot cause trouble as they need a sponsor who creates an entry barrier that needs to show some knowledge of the structure of the community. At the same time, no exploitation on a human side (single point of failure) is possible (exclusion due to personal opinions or so), as the processes are transparent and as far as possible reproducible, and a lot of sponsors exist for a lot of groups.