MINOR: Test dev cookbook deploy in PRs

[amoeba]

Fixes the dev cookbook workflow and also makes it so it will run on PRs against main. The dev cookbooks build Arrow C++ from scratch since we no longer have nightly C++ binaries.

We might consider not doing that in the future since it takes quite a long time.

The original motivation for this change was mainly because I wanted to fix the dev cookbooks so I made the workflow run on a PR since I could test but I think this has enough value to leave in for now.

[amoeba]

Getting there.

https://github.com/apache/arrow-cookbook/actions/runs/24815326450/job/72628456583?pr=400

Looks like we're picking up re2's tests. Maybe related to the recent fetchcontent upgrade. I'm trying disabling re2's tests here and will look into a fix upstream if that works.

[amoeba]

Segfault now in https://github.com/apache/arrow-cookbook/actions/runs/24817041557/job/72633422366?pr=400

Start 10: ParquetStorageServiceTest.TestCustomGrpcImpl
Errors while running CTest
10/10 Test #10: ParquetStorageServiceTest.TestCustomGrpcImpl ...***Exception: SegFault  0.55 sec
Note: Google Test filter = ParquetStorageServiceTest.TestCustomGrpcImpl

[amoeba]

The new segfault is,

[ RUN      ] ParquetStorageServiceTest.TestCustomGrpcImpl
Process 185139 stopped
* thread #21, name = 'grpcpp_sync_ser', stop reason = signal SIGSEGV: address not mapped to object (fault address: 0x8)
    frame #0: 0x00007ffff5a92a31 libarrow_flight.so.2500`grpc_closure_list_append(closure_list=0x0000000000000008, closure=0x00007fffdc0055c8) at closure.h:228:35
   225      return false;
   226    }
   227    closure->next_data.next = nullptr;
-> 228    bool was_empty = (closure_list->head == nullptr);
   229    if (was_empty) {
   230      closure_list->head = closure;
   231    } else {
(lldb) bt all
error: libarrow.so.2500 0x02f75fef: DW_TAG_member '__size' refers to type 0x0000000002f76008 which extends beyond the bounds of 0x02f75fd6
error: libarrow.so.2500 0x02f7603d: DW_TAG_member '__size' refers to type 0x0000000002f76056 which extends beyond the bounds of 0x02f76024
  thread #1, name = 'flight'
    frame #0: 0x00007fffef0a59ee libc.so.6`___lldb_unnamed_symbol3730 + 46
    frame #1: 0x00007fffef09a668 libc.so.6`___lldb_unnamed_symbol3654 + 72
    frame #2: 0x00007fffef09a6ad libc.so.6`___lldb_unnamed_symbol3655 + 13
    frame #3: 0x00007fffef11ba6d libc.so.6`epoll_wait + 29
    frame #4: 0x0000555556562f7a flight`do_epoll_wait(ps=0x000055555778b948, deadline=(millis_ = 9223372036854775807)) at ev_epoll1_linux.cc:735:19
    frame #5: 0x0000555556564971 flight`pollset_work(ps=0x000055555778b948, worker_hdl=0x00007fffffffd328, deadline=(millis_ = 9223372036854775807)) at ev_epoll1_linux.cc:1035:41
    frame #6: 0x00005555561263c7 flight`pollset_work(pollset=0x000055555778b948, worker=0x00007fffffffd328, deadline=(millis_ = 9223372036854775807)) at ev_posix.cc:242:61
    frame #7: 0x000055555612f386 flight`grpc_pollset_work(pollset=0x000055555778b948, worker=0x00007fffffffd328, deadline=(millis_ = 9223372036854775807)) at pollset.cc:48:59
    frame #8: 0x00005555559b59c1 flight`(anonymous namespace)::maybe_event_engine_pollset_work(pollset=0x000055555778b948, worker=0x00007fffffffd328, deadline=(millis_ = 9223372036854775807)) at completion_queue.cc:220:55
    frame #9: 0x00005555559b9dcc flight`cq_pluck(cq=0x000055555778b750, tag=0x00007fffffffd630, deadline=(tv_sec = 9223372036854775807, tv_nsec = 0, clock_type = GPR_CLOCK_REALTIME), reserved=0x0000000000000000) at completion_queue.cc:1312:78
    frame #10: 0x00005555559ba273 flight`grpc_completion_queue_pluck(cq=0x000055555778b750, tag=0x00007fffffffd630, deadline=(tv_sec = 9223372036854775807, tv_nsec = 0, clock_type = GPR_CLOCK_REALTIME), reserved=0x0000000000000000) at completion_queue.cc:1337:22
    frame #11: 0x0000555555690e7e flight`grpc::CompletionQueue::Pluck(this=0x00007fffffffd5f0, tag=0x00007fffffffd630) at completion_queue.h:323:44
    frame #12: 0x0000555555699ff9 flight`grpc::internal::BlockingUnaryCallImpl<google::protobuf::MessageLite, google::protobuf::MessageLite>::BlockingUnaryCallImpl(this=0x00007fffffffd990, channel=0x000055555778e4c0, method=0x00005555577685a8, context=0x00007fffffffdb10, request=0x00007fffffffdaf0, result=0x00007fffffffdad0) at client_unary_call.h:80:13
    frame #13: 0x000055555569861f flight`grpc::Status grpc::internal::BlockingUnaryCall<HelloRequest, HelloResponse, google::protobuf::MessageLite, google::protobuf::MessageLite>(channel=0x000055555778e4c0, method=0x00005555577685a8, context=0x00007fffffffdb10, request=0x00007fffffffdaf0, result=0x00007fffffffdad0) at client_unary_call.h:49:10
    frame #14: 0x000055555568da3a flight`HelloWorldService::Stub::SayHello(this=0x0000555557768580, context=0x00007fffffffdb10, request=0x00007fffffffdaf0, response=0x00007fffffffdad0) at helloworld.grpc.pb.cc:39:205
    frame #15: 0x000055555565f65f flight`TestCustomGrpcImpl() at flight.cc:388:68

[amoeba]

Thanks for looking @lidavidm. PR body updated and merged.