Floating GVariants set in GDBus skeletons leaking?

From #621 (comment 1522825):

==4763== 104 (48 direct, 56 indirect) bytes in 1 blocks are definitely lost in loss record 5,823 of 6,408
==4763==    at 0x48952CC: malloc (vg_replace_malloc.c:381)
==4763==    by 0x52F3EF3: g_malloc (gmem.c:127)
==4763==    by 0x530CD27: g_slice_alloc (gslice.c:1074)
==4763==    by 0x532CB2B: UnknownInlinedFun (gvariant-core.c:488)
==4763==    by 0x532CB2B: UnknownInlinedFun (gvariant-core.c:626)
==4763==    by 0x532CB2B: g_variant_builder_end (gvariant.c:3757)
==4763==    by 0x495C4C3: mm_bearer_ip_config_get_dictionary (mm-bearer-ip-config.c:326)
==4763==    by 0x18C0FF: UnknownInlinedFun (mm-base-bearer.c:527)
==4763==    by 0x18C0FF: bearer_update_status (mm-base-bearer.c:555)
==4763==    by 0x18C54F: disconnect_ready.lto_priv.0 (mm-base-bearer.c:1234)
==4763==    by 0x50A82CB: g_task_return_now (gtask.c:1232)
==4763==    by 0x50A854B: UnknownInlinedFun (gtask.c:1301)
==4763==    by 0x50A854B: g_task_return (gtask.c:1258)
==4763==    by 0x216477: disconnect_context_step (mm-bearer-qmi.c:2561)
==4763==    by 0x21926B: stop_network_ready (mm-bearer-qmi.c:2480)
==4763==    by 0x50A82CB: g_task_return_now (gtask.c:1232)
==4763== 
==4763== 192 (96 direct, 96 indirect) bytes in 2 blocks are definitely lost in loss record 6,154 of 6,408
==4763==    at 0x48952CC: malloc (vg_replace_malloc.c:381)
==4763==    by 0x52F3EF3: g_malloc (gmem.c:127)
==4763==    by 0x530CD27: g_slice_alloc (gslice.c:1074)
==4763==    by 0x532CB2B: UnknownInlinedFun (gvariant-core.c:488)
==4763==    by 0x532CB2B: UnknownInlinedFun (gvariant-core.c:626)
==4763==    by 0x532CB2B: g_variant_builder_end (gvariant.c:3757)
==4763==    by 0x5330A57: g_variant_valist_new (gvariant.c:5286)
==4763==    by 0x53310E3: g_variant_new_va (gvariant.c:5459)
==4763==    by 0x5331233: g_variant_new (gvariant.c:5394)
==4763==    by 0x1A30FB: update_signal_quality.lto_priv.0 (mm-iface-modem.c:1588)
==4763==    by 0x1A4BBF: load_signal_quality_ready (mm-iface-modem.c:1701)
==4763==    by 0x50A82CB: g_task_return_now (gtask.c:1232)
==4763==    by 0x50A854B: UnknownInlinedFun (gtask.c:1301)
==4763==    by 0x50A854B: g_task_return (gtask.c:1258)
==4763==    by 0x218137: get_signal_info_ready (mm-broadband-modem-qmi.c:1677)

Looks like the floating GVariants are passed into mm_gdbus_modem_set_XX() methods and those end up leaking? This is the first thing I've seen this, I wonder if it is related to some recent GDBus change.