[deepakgudiseva-nexthop]

Pre-submission checklist

• I've ran the linters locally and fixed lint errors related to the files I modified in this PR. You can install the linters by running pip install -r requirements-dev.txt && pre-commit install
• pre-commit run

Summary

addUnicastRoutes and syncFib hold every requested route twice until the hardware update returns: the thrift request and the RIB's converted copy. RibRouteTables::update then copies the converted routes once more to resolve named next-hop groups. At high ECMP width each route carries a large next-hop list, so a full-table sync or a large replace peaks at several copies of the table in sw_agent.

This moves the request routes into RouteUpdateWrapper, frees each thrift route once the RIB has converted it, resolves named next-hop groups in place on the owned vector, and frees that vector before updateFib, where the push to the hardware peaks. Existing callers are unchanged: the new addRoute and RoutingInformationBase::update overloads take rvalues, and RibRouteTables::update takes its routes by value.

Independent of #1906 (no shared files).

Test Plan

• Agent unit tests pass with this change in our fork's CI. Rebased onto main: the only difference is that the in-place resolvedRoutes line now sits in RibRouteTables::update's updated lambda.
• pre-commit run (clang-format 21.1.2) clean on the modified files.

[deepakgudiseva-nexthop]

Memory impact (#1906 + #1907)

Setup: 128-way ECMP, 22k routes (~2.8M paths), with enable_nexthop_id_manager, resolve_nexthops_from_id and drop_inline_nexthops_with_id enabled.

fboss_sw_agent memory, GiB:

	drop inline only	+ #1906 + #1907
Peak, route learn	5.8	2.2
Peak, route withdraw	5.6	2.1
Steady state	0.48	0.46

Together the two PRs cut sw_agent's peak during route updates by about 3.6 GiB (~60%). Steady state stays the same.

[deepakgudiseva-nexthop]

/autorebase