Hi
Thanks for reply.
Junos uses inet.0 for rpf check by default. Inet.2 is empty on all routers. This can be verified from my post:
admin@router# run show multicast rpf 1.1.1.1 logical-system R1
Multicast RPF table: inet.0 , 84 entries
1.1.0.0/16
Protocol: BGP
Interface: lt-1/2/0.10
It seems RPF check is unable to use the LSP route. I did a little trick with BGP policy to avoid the LSP route to mcast source (1.1.1.1) and again the multicast started working. This is probably working as it is supposed to, it's just me not understanding why. Or maybe running pim over mpls core is something very stupid from the beginning 
Luckily this is in the lab...
I also checked loopback interface is added to pim.
admin@router# run show route 1.1.1.1 logical-system R1
inet.0: 85 destinations, 186 routes (85 active, 0 holddown, 0 hidden)
Restart Complete
+ = Active Route, - = Last Active, * = Both
1.1.0.0/16 *[BGP/170] 00:00:56, localpref 100, from 10.200.90.15
AS path: 65530 I
> to 192.168.86.20 via lt-1/2/0.10
[BGP/170] 00:00:55, localpref 100, from 10.200.90.14
AS path: 65530 I
> to 192.168.86.20 via lt-1/2/0.10
[BGP/170] 00:00:56, localpref 100, from 10.200.90.16
AS path: 65530 I
> to 192.168.86.20 via lt-1/2/0.10
admin@router# run show pim join logical-system R1 239.100.100.2 extensive
Instance: PIM.master Family: INET
R = Rendezvous Point Tree, S = Sparse, W = Wildcard
Group: 239.100.100.2
Source: *
RP: 10.200.90.100
Flags: sparse,rptree,wildcard
Upstream interface: Local
Upstream neighbor: Local
Upstream state: Local RP
Uptime: 1d 23:52:02
Downstream neighbors:
Interface: fe-0/3/0.10
192.168.86.1 State: Join Flags: SRW Timeout: 166
Uptime: 1d 23:52:02 Time since last Join: 00:00:44
Group: 239.100.100.2
Source: 1.1.1.1
Flags: sparse
Upstream interface: lt-1/2/0.10
Upstream neighbor: 192.168.86.20
Upstream state: None, Local RP
Keepalive timeout: 196
Uptime: 00:19:02
Downstream neighbors:
Interface: fe-0/3/0.10 (pruned)
192.168.86.1 State: Prune Flags: SR Timeout: 166
Uptime: 00:03:53 Time since last Prune: 00:00:44
