Simplified allocation of heap buffer during gateway fetch on win32.

[Dimi1010]

• Replaced the buffer and buffer2 raw memory allocations with a managed vector.
• Removed the need of buffer2 via resizing the vector instead.
• GetAdaptersInfo will no longer be called twice if the first call does not experience buffer overflow.

[seladb]

@Dimi1010 did you have a chance to test it on Windows? We don't have tests for it unfortunately because it's not easy to run it in CI...

[Dimi1010]

@Dimi1010 did you have a chance to test it on Windows? We don't have tests for it unfortunately because it's not easy to run it in CI...

I see. Haven't tested it yet, thought it was part of the CI. Will test it after I fix the merge conflicts.

[tigercosmos]

@Dimi1010 did you have a chance to test it on Windows? We don't have tests for it unfortunately because it's not easy to run it in CI...

I am curious about why it is impossible to run it on CI?

[Dimi1010]

@seladb Tested it on Win10, using local adapter. All tests passed, except TestRawSockets which received a non-critical failure.

[seladb]

@Dimi1010 @tigercosmos I apologize, we do have tests for it which we run in CI as well:

PcapPlusPlus/Tests/Pcap++Test/Tests/LiveDeviceTests.cpp

Line 164 in 6d17ca8

PTF_ASSERT_NOT_EQUAL(defaultGateway, pcpp::IPv4Address::Zero);

I'm sorry for the confusion