flm serve crash on linux

(base) jfowers@hp-omni-1:~$ flm serve lfm2.5-it:1.2b
[FLM]  Using environment-specified port: 52625
[FLM]  Loading model: /home/jfowers/.config/flm/models/LFM2.5-1.2B-NPU2
[FLM]  Starting server on port 52625...
[FLM]  WebServer started on port 52625 with 10 I/O threads
[FLM]  Enter 'exit' to stop the server: 
[🔗 ]  TCP connection established - Remote: 127.0.0.1:35286
[TCP]  Read 687 bytes from socket
================================================
[⬇️ ]  Incoming Request: GET
[LOG]  Time stamp: 07:50:25 02:28:2026
[LOG]  Target: /
[LOG]  Version: 11
[LOG]  Keep-Alive: 1
================================================
[🔒 ]  TCP connection closed - Remote: 127.0.0.1:35286
[🔗 ]  TCP connection established - Remote: 127.0.0.1:38592
[TCP]  Read 505 bytes from socket
[✈️ ]  Handling Preflight (OPTIONS) request
Segmentation fault (core dumped)

Crash info:

 ⎿  === ExecutablePath ===
     /opt/fastflowlm/bin/flm
     === Date ===
     Sat Feb 28 07:51:55 2026
     === Title ===
     flm crashed with SIGSEGV in boost::beast::buffers_suffix<boost::beast::detail::buffers_ref<boost::beast::buffers_cat_view<boost::asio::const_buffer, boost::asio::const_buffer, boost::asio::const_buffer,
     boost::beast::http::basic_fields<std::allocator<char> >::writer::field_range, boost::beast::http::chunk_crlf> > >::consume()
     === Signal ===
     11
     === SignalName ===
     SIGSEGV
     === SegvReason ===
     reading unknown VMA
     === SegvAnalysis ===
     Segfault happened at: 0x5f406f0bb3dc <_ZN5boost5beast14buffers_suffixINS0_6detail11buffers_refINS0_16buffers_cat_viewIJNS_4asio12const_bufferES6_S6_NS0_4http12basic_fieldsISaIcEE6writer11field_rangeENS7_1
     0chunk_crlfEEEEEEE7consumeEm+204>: movzwl 0x30(%rax),%edx
     PC (0x5f406f0bb3dc) ok
     source "0x30(%rax)" (0x7ec13c031) not located in a known VMA region (needed readable region)!
     destination "%edx" ok
     Stack memory exhausted (SP below stack segment)
     === ProcCmdline ===
     flm serve lfm2.5-it:1.2b
     === StacktraceTop ===
     boost::beast::buffers_suffix<boost::beast::detail::buffers_ref<boost::beast::buffers_cat_view<boost::asio::const_buffer, boost::asio::const_buffer, boost::asio::const_buffer,
     boost::beast::http::basic_fields<std::allocator<char> >::writer::field_range, boost::beast::http::chunk_crlf> > >::consume (this=this@entry=0x7ec13c0017b0, amount=148) at
     /usr/include/boost/beast/http/impl/fields.hpp:285
     boost::beast::http::serializer<false, boost::beast::http::empty_body, boost::beast::http::basic_fields<std::allocator<char> > >::consume (this=0x7ec13c001730, n=<optimized out>) at
     /usr/include/c++/15/new:232
     boost::beast::http::detail::write_some_op<boost::beast::http::detail::write_op<boost::beast::http::detail::write_msg_op<HttpSession::handle_request(bool)::<lambda(boost::beast::error_code, std::size_t)>,
     boost::asio::basic_stream_socket<boost::asio::ip::tcp, boost::asio::any_io_executor>, false, boost::beast::http::empty_body, boost::beast::http::basic_fields<std::allocator<char> > >,
     boost::asio::basic_stream_socket<boost::asio::ip::tcp, boost::asio::any_io_executor>, boost::beast::http::detail::serializer_is_done, false, boost::beast::http::empty_body,
     boost::beast::http::basic_fields<std::allocator<char> > >, boost::asio::basic_stream_socket<boost::asio::ip::tcp, boost::asio::any_io_executor>, false, boost::beast::http::empty_body,
     boost::beast::http::basic_fields<std::allocator<char> > >::operator() (this=0x7ec151bf58a0, ec=..., bytes_transferred=197) at /usr/include/boost/beast/http/impl/write.hpp:131
     boost::asio::detail::binder2<boost::beast::http::detail::write_some_op<boost::beast::http::detail::write_op<boost::beast::http::detail::write_msg_op<HttpSession::handle_request(bool)::<lambda(boost::beast
     ::error_code, std::size_t)>, boost::asio::basic_stream_socket<boost::asio::ip::tcp, boost::asio::any_io_executor>, false, boost::beast::http::empty_body,
     boost::beast::http::basic_fields<std::allocator<char> > >, boost::asio::basic_stream_socket<boost::asio::ip::tcp, boost::asio::any_io_executor>, boost::beast::http::detail::serializer_is_done, false,
     boost::beast::http::empty_body, boost::beast::http::basic_fields<std::allocator<char> > >, boost::asio::basic_stream_socket<boost::asio::ip::tcp, boost::asio::any_io_executor>, false,
     boost::beast::http::empty_body, boost::beast::http::basic_fields<std::allocator<char> > >, boost::system::error_code, long unsigned int>::operator() (this=0x7ec151bf58a0) at
     /usr/include/boost/asio/detail/bind_handler.hpp:181
     boost::asio::detail::binder0<boost::asio::detail::binder2<boost::beast::http::detail::write_some_op<boost::beast::http::detail::write_op<boost::beast::http::detail::write_msg_op<HttpSession::handle_reques
     t(bool)::<lambda(boost::beast::error_code, std::size_t)>, boost::asio::basic_stream_socket<boost::asio::ip::tcp, boost::asio::any_io_executor>, false, boost::beast::http::empty_body,
     boost::beast::http::basic_fields<std::allocator<char> > >, boost::asio::basic_stream_socket<boost::asio::ip::tcp, boost::asio::any_io_executor>, boost::beast::http::detail::serializer_is_done, false,
     boost::beast::http::empty_body, boost::beast::http::basic_fields<std::allocator<char> > >, boost::asio::basic_stream_socket<boost::asio::ip::tcp, boost::asio::any_io_executor>, false,
     boost::beast::http::empty_body, boost::beast::http::basic_fields<std::allocator<char> > >, boost::system::error_code, long unsigned int> >::operator() (this=0x7ec151bf58a0) at
     /usr/include/boost/asio/detail/bind_handler.hpp:56

cc @superm1

Start by reproducing the crash with flm serve lfm2.5-it:1.2b and an OPTIONS request, then inspect HttpSession::handle_request using the Boost.Beast stack trace as the entry point. Done means the preflight request no longer causes a SIGSEGV and the server remains running.