Fix cppcheck [useInitializationList] warnings - #191

I don't understand you, this PR isn't related to #190, please review here this PR.

In short, the problem is: you're fixing warnings not code.

Some examples:

• in src/FileLock.h you move the initialization of hd into the initialization list of the constructor. You didn't notice that it was a false positive from cppcheck, because hd is of type HANDLE which actually is a void *. Thus the change doesn't come with any benefit but reduced readability.

• still staying at src/FileLock.h, there is a #if block starting inside a function and spanning through to the end of the file. That's a really really bad thing and should have been fixed first.

• in src/IP2Country.cpp you move m_DataBaseName but not m_DataBasePath which has the same type (wxString). I admit it has a more complex initializer, but that can be simplified to allow it going, too.

• moving POD-types into the initialization list doesn't do any benefit (nor harm, to be honest). I guess the warnings you got for src/webserver/src/WebSocket.cpp are because cppcheck didn't know what uint32 was.

• and last but not least, I miss the style from your patch. You pick some variables (sometimes seemingly randomly) to move and leave the others. With the exceptions being src/utils/wxCas/src/onlinesig.cpp and the ones where there were only one member to initialize...

If you're already using cppcheck, there are more severe issues in the code.... (just a hint). If I were you I would certainly open a branch and commit there all the changes required by cppcheck, all the changes that were made on the way to fix bugs you encounter meanwhile. And maybe create a configuration file for cppcheck to prevent false positives. And, while we're already at it, use meaningful commit messages, describing what and why you change.

This PR should be too old for the current code.
It is free to download cppcheck and run its GUI.
For me, there were a few useful findings, including real bugs.
Expect a few false positives too, but with small experience those could be easily discarded.

Thanks for this PR — the core idea is correct and worth landing. I rebased it onto the current master and resolved the conflicts by hand. Here's what changed:

Applied as-is (clean auto-merge):

• src/AsyncDNS.cpp — m_ipName moved to init list, preserving the thread-safety comment
• src/GapList.h — m_it in const_iterator
• src/Tag.cpp — m_Name in both CTag(wxString) and CTag(CTag&)
• src/TextClient.cpp — nHash in SearchFile
• src/utils/wxCas/src/onlinesig.cpp — all 7 members (this one was the cleanest in the original PR)

Applied with fixes (conflict resolution):

• src/CanceledFileList.cpp — moved to init list but removed wxT(), which was retired tree-wide in the meantime
• src/Friend.cpp — moved m_UserHash to init list; kept m_HasFriendSlot = false in the body (that field was added after this PR was opened, in GUI: friend slot keeps disabling itself #282)
• src/webserver/src/WebSocket.cpp — kept the init-list entries, but preserved the [4096+1] allocation and its comment (added in Off-by-one heap write in CWebSocket::OnReceive #873 to fix an off-by-one); removed the now-redundant body assignments

Dropped intentionally:

• src/FileLock.h — HANDLE is void* (a POD type), so cppcheck's warning here is a false positive; moving a multi-line CreateFileA(...) call into the init list buys nothing and hurts readability
• src/IP2Country.cpp — this file was completely rewritten to use libmaxminddb instead of libGeoIP; the GeoIP.dat initializer no longer exists

@got3nks , what do you think? Harmless?

The FileLock.h #if inside a function issue can be handled elsewhere.

Looks safe to land — sanity-checked your four decisions against current master and they all hold up:

• Applied as-is — AsyncDNS, GapList, Tag, TextClient, onlinesig: assignment-in-body → init-list is purely syntactic, no runtime effect.
• Applied with fixes — CanceledFileList (drop wxT()), Friend (keep m_HasFriendSlot = false from GUI: friend slot keeps disabling itself #282), WebSocket (preserve [4096+1] from Off-by-one heap write in CWebSocket::OnReceive #873): all correct against the current files; the surrounding code you kept is the bit that actually matters.
• Dropped — FileLock.h's HANDLE is POD void* so cppcheck's warning is a false positive, and IP2Country.cpp is now libmaxminddb-based with no GeoIP.dat initializer to touch. Both N/A.

Agree on FileLock.h — the per-platform #if-split inside the class body (two constructors, two destructors, two private: sections with different members) is its own much bigger refactor target and shouldn't bottle this PR.

Harmless and worth landing.